Your SlideShare is downloading. ×
0205 f01 international research roadmap
Upcoming SlideShare
Loading in...5

Thanks for flagging this SlideShare!

Oops! An error has occurred.

Saving this for later? Get the SlideShare app to save on your phone or tablet. Read anywhere, anytime – even offline.
Text the download link to your phone
Standard text messaging rates apply

0205 f01 international research roadmap


Published on

Published in: Technology, News & Politics
1 Comment
1 Like
No Downloads
Total Views
On Slideshare
From Embeds
Number of Embeds
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

No notes for slide


  • 1.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     ICT  Seventh  Framework  Programme  (ICT  FP7)       Grant  Agreement  No:  288828   Bridging  Communities  for  Next  Generation  Policy-­‐Making         Towards  Policy-­‐making  2.0:   The  International  Research  Roadmap  on     ICT  for  Governance  and  Policy  Modelling       Internal  Deliverable  Form   Project  Reference  No.   ICT  FP7  288828   Deliverable  No.     D2.2.2   Relevant  Workpackage:   WP2   Nature:   Report   Dissemination  Level:   Public   Document  version:   FINAL  1.0   Date:   12/09/2013   Authors:   David   Osimo   &   Francesco   Mureddu   (T4I2),   Riccardo   Onori   &   Stefano  Armenia  (CATTID),  Gianluca  Carlo  Misuraca  (IPTS)   Reviewers:   Eva  Jaho  (ATC),  Andrea  Bassi  (MI)   Document  description:   This   deliverable   describes   the   final   version   of   the   new   International   Research   Roadmap   on   ICT   Tools   for   Governance   and   Policy  
  • 2.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   Modelling           History   Version   Date   Reason   Revised  by   1.0   30/06/2013   1st  draft   T4I2   2.0   12/07/2013   2nd  draft  sent  for  peer   T4I2   review       26/07/2013   Peer   review   feedback   3.0   09/08/2013   3rd   draft   sent   for   final   T4I2   confirmation     06/09/2013   Partners’  approval   1.0   12/09/2013   Final   version   sent   to   ATC   the  PO  and  reviewers   and   ATC,  MI   ATC,   DIAG,   IPTS,  MI   W3C,   2  |  P a g e  
  • 3.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   TABLE  OF  CONTENTS     EXECUTIVE  SUMMARY ................................................................................................................................... 5   1.   BACKGROUND:  WHY  A  ROADMAP?........................................................................................................ 8   1.1.   1.2.   1.3.   1.4.   The  rationale  of  the  roadmap:  what  is  the  problem? ............................................................................. 8   An  open  and  recursive  methodology ...................................................................................................... 9   Scope  and  definition.............................................................................................................................. 16   Policy:  Between  politics  and  services .................................................................................................... 19   2.   NOT  JUST  ANOTHER  HYPE:  THE  DEMAND  SIDE  OF  POLICY-­‐MAKING  2.0 ................................................ 20   2.1.   The  typical  tasks  of  policy-­‐makers:  the  policy  cycle .............................................................................. 21   2.2.   The  traditional  tools  of  policy-­‐making................................................................................................... 22   2.3.   The  key  challenges  of  policy-­‐makers ..................................................................................................... 23   2.3.1.   Detect  and  understand  problems  before  they  become  unsolvable............................................... 24   2.3.2.   Generate  high  involvement  of  citizens  in  policy-­‐making................................................................ 24   2.3.3.   Identify  “good  ideas”  and  innovative  solutions  to  long-­‐standing  problems .................................. 24   2.3.4.   Reduce  uncertainty  on  the  possible  impacts  of  policies ................................................................ 25   2.3.5.   Ensure  long  -­‐  term  thinking ............................................................................................................ 27   2.3.6.   Encourage  behavioural  change  and  uptake ................................................................................... 27   2.3.7.   Manage  crisis  and  the  “unknown  unknown” ................................................................................. 27   2.3.8.   Moving  from  conversations  to  action ............................................................................................ 28   2.3.9.   Detect  non-­‐compliance  and  mis-­‐spending  through  better  transparency ...................................... 28   2.3.10.   Understand  the  impact  of  policies ............................................................................................... 29   2.4.   When  policy-­‐making  2.0  becomes  a  reality:  a  tentative  vision  for  2030............................................... 29   2.4.1.   Agenda  setting  phase:  recognizing  the  problem ............................................................................ 29   2.4.2.   Policy  design ................................................................................................................................... 30   2.4.3.   Implementation.............................................................................................................................. 31   2.4.4.   Evaluation ....................................................................................................................................... 31   2.5.   The  key  challenges  for  policy  makers  and  the  corresponding  phases  in  the  policy  cycle ..................... 32   3.   THE  SUPPLY  SIDE:  CURRENT  STATUS  AND  THE  RESEARCH  CHALLENGES................................................ 33   3.1.   Policy  Modelling .................................................................................................................................... 33   3.1.1.   Systems  of  Atomized  Models ......................................................................................................... 33   3.1.2.   Collaborative  Modelling ................................................................................................................. 42   3.1.3.   Easy  Access  to  Information  and  Knowledge  Creation .................................................................... 53   3.1.4.   Model  Validation ............................................................................................................................ 56   3.1.5.   Immersive  Simulation..................................................................................................................... 59   3.1.6.   Output  Analysis  and  Knowledge  Synthesis..................................................................................... 61   3.2.   Data-­‐powered  Collaborative  Governance ............................................................................................. 64   3.2.1.   Big  Data .......................................................................................................................................... 64   3.2.2.   Opinion  Mining  and  Sentiment  Analysis......................................................................................... 78   3.2.3.   Visual  Analytics  for  collaborative  governance:  the  opportunities  and  the  research  challenges.... 85   3.2.4.   Serious  Gaming  for  Behavioural  Change ........................................................................................ 98   3.2.5.   Linked  Open  Government  Data .................................................................................................... 103   3.2.6.   Collaborative  Governance ............................................................................................................ 109   3.2.7.   Participatory  Sensing .................................................................................................................... 113   3.2.8.   Identity  Management................................................................................................................... 117   3.2.9.   Global  Systems  Science ................................................................................................................ 120   4.   THE  CASE  FOR  POLICY-­‐MAKING  2.0:  EVALUATING  THE  IMPACT .......................................................... 127   4.1.   Cross  analysis  of  case  studies .............................................................................................................. 127   4.1.1.   Global  Epidemic  and  Mobility  Model ........................................................................................... 128   Impact  of  Gleam ......................................................................................................................................... 128   4.1.2.   UrbanSim ...................................................................................................................................... 129   3  |  P a g e  
  • 4.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   4.1.3.   Opinion  Space............................................................................................................................... 130   4.1.4.   2050  Pathways  Analysis................................................................................................................ 132   4.1.5.   Cross  analysis  of  the  case  studies................................................................................................. 134   4.2.   Survey  of  Users’  needs  results............................................................................................................. 136   4.3.   Analysis  of  the  prize  winners............................................................................................................... 139   4.4.   Lessons  learnt  from  cases  and  prize.................................................................................................... 143   4.5.   An  additional  research  challenge:  counterfactual  impact  evaluation  of  Policy  Making  2.0................ 144   5.   CONCLUSIONS:  POLICY-­‐MAKING  2.0  BETWEEN  HYPE  AND  REALITY .................................................... 149   6.   REFERENCES ....................................................................................................................................... 153   7.   LIST  OF  ACRONYMS ............................................................................................................................ 157             LIST  OF  FIGURES   Figure  1:  the  fragmentation  of  policy-­‐making  2.0.................................................................................................. 8   Figure  2  Outline  of  the  participatory  process ...................................................................................................... 10   Figure  3:  Policy  Cycle  and  Related  Activities ........................................................................................................ 22   Figure  4:  Total  Disasters  Reported ...................................................................................................................... 28   Figure  5:  Agricultural  Production  and  Externalities  Simulator  (APES) ............................................................... 36   Figure  6:  Conversational  Modelling  Interface .................................................................................................... 45   Figure  7:  the  PADGET  Framework ....................................................................................................................... 46   Figure  8:  the  Time-­‐Space  Matrix ......................................................................................................................... 49   Figure  9:  COMA,  COllaborative  Modelling  Architecture .................................................................................... 50   Figure  10:  OCOPOMO  eParticipation  Platform................................................................................................... 51   Figure  11:  Twitrratr.............................................................................................................................................. 81   Figure  12:  Wordclouds......................................................................................................................................... 82   Figure  13:  UserVoice............................................................................................................................................ 82   Figure  14    Open  Data  Business  Model  (source:  Istituto  Superiore  Mario  Boella) .............................................. 106   Figure  15  -­‐LOD  providers  and  their  linkages ...................................................................................................... 107   Figure  16  Rating  other  opinions'  in  Opinion  Space ............................................................................................ 131   Figure  17  Playing  the  My2050  game  for  the  demand  side................................................................................. 133   Figure   18   Adoption   of   ICT   Tools   and   Methodologies   for   policy-­‐making   (source:   CROSSOVER   Survey   of   Users’   Needs  2012) ....................................................................................................................................................... 137   Figure   19   Needs   and   Challenges   in   the   Policy   Making   Process   (source:   CROSSOVER   Survey   of   Users’   Needs   2012) .................................................................................................................................................................. 138   Figure  20:  a  proposed  evaluation  framework  for  policy-­‐making  2.0 ................................................................. 144   Figure  21:  Relation  Between  Policy-­‐Making  Needs  and  Research  Challenges................................................... 149     4  |  P a g e  
  • 5.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   Executive  Summary   This   deliverable   introduces   and   describes   the   interim   version   of   the   new   International   Research   Roadmap  on  ICT  tools  for  Governance  and  Policy  Modelling,  renamed  by  the  project  team  as  “Policy-­‐ Making   2.0”,   one   of   the   core   outputs   of   the   Crossover   project,   which   is   developed   under   WP2   Content  Production.     The   roadmap   aims   to   establish   the   scientific   and   political   basis   for   long-­‐lasting   interest   and   commitment   to   next   generation   policy-­‐making   by   researchers   and   policy-­‐makers.   In   doing   so,   it   contains  an  analysis  of  what  technologies  are  currently  available,   for  what  concrete  purposes,   and   what  could  become  available  in  the  future.  The  main  rationale  for  such  a  document  is  the  current   fragmentation   of   the   landscape   between   different   stakeholders,   disciplines,   policy   domains   and   geographical  areas.     The  document  is  the  result  of  a  highly  participative  process  undergone  between  the  first  draft  and   the   final   roadmap,   with   the   involvement   of   hundreds   of   people   through   11   different   input   methods,   from  live  workshops  to  online  discussion.     5  |  P a g e  
  • 6.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   After  a  brief  introduction  of  the  background,  the  document  analyses  the  demand  side:  the  current   status   of   policy-­‐making,   with   the   key   tasks   (illustrated   by   the   traditional   policy   cycle)   and   existing   challenges:   a. Detect  and  understand  problems  before  they  become  unsolvable b. Generate  high  involvement  of  citizens  in  policy-­‐making c. Identify  “good  ideas”  and  innovative  solutions  to  long-­‐standing  problems d. Reduce  uncertainty  on  the  possible  impacts  of  policies e. Ensure  long  -­‐  term  thinking f. Encourage  behavioural  change  and  uptake g. Manage  crisis  and  the  “unknown  unknown” h. Moving  from  conversations  to  action i. Detect  non-­‐compliance  and  mis-­‐spending  through  better  transparency j. Understand  the  impact  of  policies It   then   presents   a   concrete   tentative   vision   of   how   policy-­‐making   could   look   in   2030,   if   these   challenges  were  overcome.   Section   3   represents   the   core   of   the   roadmap   and   presents   the   key   research   challenges   to   be   addressed   to   achieve   this   vision,   updating   the   original   version   based   on   the   input   of   the   consultation.  For  each  research  challenge,  it  presents  the  current  status,  the  existing  gaps,  and  short   and  long  term  research  perspectives.  The  key  research  challenges  are:   1. Policy  Modelling 1.1. Systems  of  Atomized  Models 1.2. Collaborative  Modelling 1.3. Easy  Access  to  Information  and  Knowledge  Creation 1.4. Model  Validation 1.5. Immersive  Simulation 1.6. Output  Analysis  and  Knowledge  Synthesis 2. Data-­‐powered  Collaborative  Governance 2.1. Big  Data 2.2. Opinion  Mining  and  Sentiment  Analysis 2.3. Visual  Analytics  for  collaborative  governance:  the  opportunities  and  the  research  challenges 2.4. Serious  Gaming  for  Behavioural  Change 2.5. Linked  Open  Government  Data 2.6. Collaborative  Governance 2.7. Participatory  Sensing 2.8. Identity  Management 2.9. Global  Systems  Science   But   to   what   extent   policy-­‐making   2.0   can   be   said   to   genuinely   improve   policy-­‐making?   Section   4   looks  at  the  available  evidence  about  the  impact  of  policy-­‐making  2.0,  across  case  studies,  the  survey   and  the  prize.  As  it  emerges  that  no  robust  impact  evaluation  is  available,  we  propose  an  additional   6  |  P a g e  
  • 7.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   research   challenge   on   impact   evaluation   of   policy-­‐making   accompanied   by   a   proposed   evaluation   framework.     Finally,   we   summarize   the   findings   of   the   document   bringing   together   the   different   sections,   suggesting   that   policy-­‐making   2.0   cannot   be   considered   the   panacea   for   all   issues   related   to   bad   public   policies,   but   that   at   the   same   time   it   is   more   than   just   a   neutral   set   of   disparate   tools.   It   provides  an  integrated  and  mutually  reinforcing  set  of  methods  that  share  a  similar  vision  of  policy-­‐ making   and   that   should   be   addressed   in   an   integrated   and   strategic   way;   and   it   provides   opportunities  to  improve  the  checks  and  balances  systems  behind  decision  making  in  government,   and  as  such  it  should  be  further  pursued.       and  as  such  it  should  be  further  pursued.     7  |  P a g e  
  • 8.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   1.   BACKGROUND:  WHY  A  ROADMAP?   1.1. The  rationale  of  the  roadmap:  what  is  the  problem?     The   CROSSOVER   project   aims   to   consolidate   and   expand   the   existing   community   on   ICT   for   Governance  and  Policy  Modelling  (built  largely  within  FP7)  by:     -­‐   Bringing   together   and   reinforcing   the   links   between   the   different   global   communities   of   researchers  and  experts:  it  will  create  directories  of  experts  and  solutions,  and  animate  knowledge   exchange  across  communities  of  practice  both  offline  and  online;   -­‐   Reaching   out   and   raising   the   awareness   of   non-­‐experts   and   potential   users,   with   special   regard  to  high-­‐level  policy-­‐makers  and  policy  advisors:  it  will  produce  multimedia  content,  a  practical   handbook  and  high-­‐level  policy  conferences  with  competition  for  prizes;   -­‐   Establishing  the  scientific  and  political  basis  for  long-­‐lasting  interest  and  commitment  to  next   generation   policy-­‐making,   beyond   the   mere   availability   of   FP7   funding:   it   will   focus   on   use   cases   and   a  demand-­‐driven  approach,  involving  policy-­‐makers  and  advisors.   The   CROSSOVER   project   pursues   this   goal   through   a   combination   of   content   production,   ad   hoc   and   well-­‐designed  online  and  offline  animation;  as  well  as  strong  links  with  existing  communities  outside   the  CROSSOVER  project  and  outside  the  realm  of  e-­‐Government.     The   present   deliverable   is   one   of   the   core   outputs   of   the   project:   the   International   Research   Roadmap  on  ICT  Tools  for  Governance  and  Policy  Modelling.  It  aims  to  create  a  common  platform   between  actors  fragmented  in  different  disciplines,  policy  domains,  organisations  and  geographical   areas,  as  illustrated  in  the  figure  below.     Figure  1:  the  fragmentation  of  policy-­‐making  2.0     But  most  of  all,  it  aims  to  provide  a  clear  outline  of  what  technologies  are  available  now  for  policy-­‐ makers  to  improve  their  work,  and  what  could  become  available  tomorrow.     8  |  P a g e  
  • 9.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   CROSSOVER   builds   on   the   results   of   the   CROSSROAD   project1,   which   elaborated   a   research   roadmap   on   the   same   topic   along   the   whole   of   2010.   With   respect   to   the   previous   roadmap,   this   document   is   firstly  a  revised  and  updated  version.  Beside  this,  it  contains  some  fundamental  novelties:   -­‐ A  demand-­‐driven  approach:  rather  than  focussing  on  the  technology,  the  present  roadmap   starts   from   the   needs   and   the   activities   of   policy-­‐making   and   then   links   the   research   challenges  to  them.     -­‐ An  additional  emphasis  on  cases  and  applications:  for  each  research  challenge,  we  indicate   relevant  cases  and  practical  solutions   -­‐ A   clearer   thematic   focus   on   ICT   for   Governance   and   Policy-­‐Modelling,   by   dropping   more   peripheral   grand   challenges   of   Government   Service   Utility   and   Scientific   Base   for   ICT-­‐ enabled  Governance   -­‐ A   global   coverage:   while   CROSSROAD   focussed   on   Europe,   CROSSOVER   includes   cases   and   experiences  from  all  over  the  world   -­‐ A   living  roadmap:   the   present   deliverable   is   accompanied   by   an   online   repositories   of   tools,   people  and  applications   1.2. An  open  and  recursive  methodology     The  present  Research  Roadmap  on  Policy-­‐Making  2.0  is  developed  with  a  sequential  approach  based   on   the   existing   research   roadmap   developed   by   the   CROSSROAD   project.   In   order   to   achieve   the   goals  of  overcoming  the  fragmentation,  an  open  and  inclusive  approach  was  necessary.   In   the   initial   phase   of   the   project,   up   to   M6   (March   2012),   the   consortium   started   a   collection   of   literature,   information   about   software   tools   and   applications   cases.   In   addition   to   this   desk-­‐based   review,   the   document   has   benefited   from   the   informal   discussions   being   held   on   the   LinkedIn   group   of  the  project  (Policy-­‐making  2.0),  where  more  than  800  practitioners  and  researchers  are  discussing   the  practices  and  the  challenges  of  policy-­‐making.   The   first   draft   of   the   roadmap   was   then   released   in   M9   (June   2012)   of   the   project,   for   public   feedback.   The   publication   of   the   deliverable   kicked   off   the   engagement   activities   of   the   project,   designed  to  provide  further  input  and  to  improve  the  roadmap:   -­‐ As   soon   as   it   was   released,   the   preliminary   version   of   the   roadmap   was   published   in   commentable   format   on   the   project   website   http://www.CROSSOVER-­‐   Animators   stimulated   discussion   about   it   and   generated   comments   by   researchers   and   practitioners  alike.  This  participatory  process  helped  enriching  the  roadmap,  which  was  then   published  in  its  final  version  after  validation  by  the  community/ies  of  practitioners  and  policy   makers   -­‐ Two   workshops   organised   by   the   project   aimed   at   gathering   input   on   the   research   challenges  and  feedback  on  the  proposed  roadmap     -­‐ An  online  survey,  as  well  as  several  focus  groups  and  meetings  with  practitioners  from  civil   society  and  government  helped  to  focus  the  roadmap  on  the  actual  needs                                                                                                                             1   9  |  P a g e  
  • 10.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     Figure  2  Outline  of  the  participatory  process   The   process   for   updating   the   roadmap   included   therefore   a   wide   set   of   contributions.   Firstly,   the   Crossroad  roadmap  was  enriched  with  desk-­‐based  research:  202  cases  collected  in  the  platform  +  4   cases  collected  and  described  in  the  case  studies  performed  by  the  National  Technical  University  of   Athens  (NTUA),  and  the  50  applications  to  the  prize.     This  first  draft  was  then  published  for  comments  by  some  of  the  800  members  of  the  LinkedIn  group   who   also   provided   relevant   cases.   An   additional   survey   of   users’   needs   provides   provided   insights   from   240   respondents   and   over   200   people   presents   presented   at   focus   groups.   Additional   discussions   with   Global   Systems   Science     community,   third   party   workshops   and   the   US   Policy   Informatics  Network    helped  in  refine  refining  further  the  roadmap.   The   two   workshops   provided   high-­‐quality   insight   that   enriched   the   roadmap   with   specific   contributions.     In   the   table   below   we   outline   in   detail   the   specific   contribution   of   each   section   of   the   roadmap,   that   is  described  in  full  in  the  following  section.   10  |  P a g e  
  • 11.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP       Type  of  contribution   Extent  of  the  contribution   Contribution  to  the  roadmap   1) Comments to the roadmap • 40  comments   • 9  different  experts   • • • • 2) Presentations in the PMOD • Papers  received:  42   • Registered  participants:  70     • No.  Countries’  citizens  present:   20   • Linked  Open  Government  Data   • 16  presentations   • 30  participants   • Collaborative  Modelling   • Systems  of  Atomized  Models   • Opinion  Mining   • Impact  of  policy  making  2.0   • Roadmap  methodology   • Linked  Open  Government  Data   • Opinion  Mining   • Collaborative  Governance workshop 3) Presentations in Transatlantic workshop 4) Survey of User’s Needs the   • 236  respondents   • 33%  engaged  in  policy  design   • 27%  engaged  in  monitoring  and   evaluation   • 22%  engaged  in  agenda  setting   • 18%  engaged  in  policy   implementation   5) Focus groups   6) Case studies 7) Analysis of the prize 8) LinkedIn group 139  attendants  -­‐  Forum  PA,  the   Italian  leading  conference  on  e-­‐ government     • 35  attendants-­‐  INSITE  event  on   sustainability     • 40  attendants  -­‐  Webinar  for  the   United  Nations  Development   Programme   • Collection  of  202  tools  and   practices   • Elicitation  of  20  best  practices   • Further  elicitation  of  4  best   practices  for  in-­‐depth  case   study   • • • • 47  submission  received   10  short  listed   3  winners   840  participants   Visual  Analytics   Systems  of  Atomized  Models   Model  Validation   Serious  Gaming     • Impact  of  policy  making  2.0   • Roadmap  methodology   • Impact  of  policy  making  2.0   • Roadmap  methodology   • Annex  with  a  repository  of  cases   • Analysis  of  the  prize  process  on  the   Impact  Chapter   • Comments  to  the  roadmap   • Increased  attendance  to  the   workshops   • Collection  of  practices  and  tools   Table  1  Contributions  to  the  roadmap   1) Comments  to  the  Roadmap   The  roadmap  has  been  published  in  commentable  format  in  two  different  versions:  a  short  one  on   Makingspeechtalk2,   and   a   full   version   (downloadable   after   answering   the   survey   on   the   needs   of                                                                                                                             2   11  |  P a g e  
  • 12.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   policy-­‐makers)   available   in   the   CROSSOVER   website3.   Everybody   was   able   to   comment   on   single   parts  of  the  roadmap  or  to  propose  new  topics,  application  cases  and  research  challenges.  The  aim   of   publishing   the   document   in   commentable   format   was   to   get   the   input   from   experts   for   co-­‐ creating   the   roadmap.   More   specifically   we   were   interested   in   knowing   if   the   current   formulation   of   the   research   challenge   was   acceptable,   and   we   wanted   to   collect   best   practices   and   application   cases  from  the  community  of  experts  and  practitioners  at  large.  As  already  mentioned,  the  roadmap   received  over  40  useful  and  detailed  comments  from  a  number  of  experts  in  the  different  domains.   2) PMOD  Workshop   The   June   2012   workshop   was   the   first   of   three   to   be   organised   under   the   CROSSOVER   project.   Formally   titled   "Using   Open   Data:   policy   modelling,   citizen   empowerment,   data   journalism"   but   generally   referred   to   by   the   term   PMOD   (policy   modelling),   it   set   out   to   explore   whether   advocates'   claims   of   the   huge   potential   for   open   data   as   an   engine   for   a   new   economy,   as   an   aid   to   transparency   and,   of   particular   relevance   to   CROSSOVER,   as   an   aid   to   evidence-­‐based   policy   modelling,   were   justified.   In   terms   of   organization,   the   event   was   run   as   a   W3C/CROSSOVER   workshop  and  held  at  the  European  Commission's  Albert  Borschette  Conference  Centre  in  the  two   days   immediately   prior   to   the   Digital   Agenda   Assembly.   That   combination   helped   to   secure   good   support  from  a  high  calibre  audience.  42  papers  were  received  and  the  majority  was  accepted  by  the   programme   committee   for   full   presentation.   Authors   of   several   other   papers   plus   members   of   the   programme  committee,  the  CROSSOVER  animators  and  a  small  number  of  invited  guests  comprised   the   70   registered   attendees   of   which   67   turned   up.   The   event   reached   a   larger   audience   through   organising   a   networking   event   on   the   evening   following   the   workshop   to   which   attendees   of   the   data   workshop   at   the   Digital   Agenda   Assembly   were   invited.   Furthermore,   through   the   live   IRC   channel   and   Tweets   using   the   #pmod   hashtag,   others   were   able   to   monitor   proceedings.   The   agenda,  attendee  list  and  final  report  are  all  available  on  the  W3C    Web  site  which  provides  a  high   profile  for  the  workshop  and  the  project.   Most  of  the  results  of  the  workshop  were  used  to  improve  the  research  challenge  on  Linked  Open   Government  Data.     3) Transatlantic  Workshop   The   Transatlantic   Research   on   Policy   Modelling   Workshop   that   was   held   in   Washington,   DC   on   January   28th   and   29th,   2013.   It   was   organized   by   the   Millennium   Institute   and   the   New   America   Foundation  (NAF),  Washington,  DC,  USA.  NAF  is  a  nonprofit,  nonpartisan  public  policy  institute  that   invests  in  new  thinkers  and  new  ideas  to  address  the  next  generation  of  challenges  facing  the  United   States.  This  event  brought  together  speakers  and  attendees  working  and/or  interested  in  improving   ICT   tools   for   education   and   policy   makers.   The   speakers   and   attendees   came   from   a   diverse   background,  both  technical  and  non-­‐technical  to  share  experiences  and  knowledge  and  discuss  ways   to  make  the  current  state  of  modelling  and  ICT  more  accessible  and  attractive  for  decision  makers   on   both   sides   of   the   Atlantic   Ocean.   The   models   presented   in   the   workshop   have   been   integrated   in   the   “Collaborative   Modelling”,   “Systems   of   Atomized   Models”   and   “Opinion   Mining”   research   challenges.     4) Survey  of  User’s  Needs                                                                                                                             3  http://www.CROSSOVER-­‐       12  |  P a g e  
  • 13.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   The   Survey   of   Users’   Needs   performed   within   the   scope   of   the   CROSSOVER   project   aimed   at   collecting   the   views   and   the   requirements   of   policy-­‐making   stakeholders.   More   in   particular   the   survey   intended   to   stimulate   actual   and   potential   practitioners,   such   as   decision   makers   (government   official   involved   in   the   policy-­‐making   process)   or   policy   advisors   (technical   expert   advising  decision-­‐makers  from  outside  government)  to  provide  input,  feedback  and  validation  to  the   new   research   roadmap   on   ICT   tools   for   Governance   and   Policy   Modelling   under   development   (CROSSOVER,   2012b).   About   450   people   took   part   in   the   overall   exercise,   combining   live   meetings   (214)   and   online   survey   (240+   answers),   providing   concrete   elements   to   improve   the   CROSSOVER   roadmap  and  the  other  activities  to  be  carried  out  by  the  project.     5) Focus  groups   In   addition   to   the   survey,   Tech4i2   ran   a   series   of   dedicated   meetings   where   the   roadmap   was   presented   and   followed   up   by   intense   dedicated   discussion.   These   events   where   all   high-­‐profile,   attended  by  policy-­‐makers  in  the  broad  sense:  not  only  government  officials,  but  also  policy  advisors   and  civil  society  organisations.  More  precisely  three  events  have  been  run:   • On  the  17th  of  May  2012  CROSSOVER  was  invited  to  give  a  keynote  speech  to  ForumPA   on   the   CROSSOVER   Research   Roadmap.   FORUM   PA   is   a   leading   European   exhibition   exploring   innovation   in   Public   Administration   and   local   systems.   For   22   years,   FORUM   PA   has   attracted   thousands   of   visitors   and   hundreds   of   exhibitors   (public   authorities,   private   companies   and   citizens)   to   come   together   and   learn   and   the   participation   of   important   leaders:   ministers,   Nobel   prize   winners   (Amartya   Sen,   Edward   Prescott),   industry  leaders  (Luca  Cordero  di  Montezemolo)  and  hundreds  of  speakers.   • On   May   24th   2012,   CROSSOVER   was   invited   to   attend   the   HUB/Insite   project   meeting   of   sustainability   practitioners   from   all   over   Europe.   The   Hub   and   the   INSITE   Project   brought  together  more  than  25  sustainability  practitioners  working  at  the  cutting  edge   of  innovation  within  industry,  urban  development,  energy,  technology  and  policy  across   Europe.  This  includes  people  tackling  today’s  key  challenges  in  carbon  reduction,  smart   cities,  governance  and  behavioural  change  across  all  these  areas.  Tech4i2  presented  the   Research   Roadmap,   and   facilitated   a   dedicated   session   CROSSOVER   was   invited   to   attend   the   HUB/Insite   project   meeting   of   sustainability   practitioners   from   all   over   Europe.     • On  March  22nd  2012,  CROSSOVER  was  invited  to  present  the  policy-­‐making  2.0  model   to   the   practitioners   of   the   “governance”   network   of   UNDP   –   Europe   and   CIS,   which   included   about   40   people   from   Central   and   Eastern   Europe.   Webinar   for   the   United   Nations  Development  Programme  –  Europe  and  CIS   6) Case  Studies   Within   the   scope   of   the   CROSSOVER   project,   the   European   Commission's   Joint   Research   Centre,   Institute  for  Prospective  Technological  Studies  (JRC-­‐IPTS),  in  collaboration  with  a  team  of  experts  of   the   National   Technical   University   of   Athens   (NTUA)   carried   out   the   activity   of   mapping   and   identification   of   Case   Studies   on   ICT   solutions   for   governance   and   policy   modelling   (CROSSOVER,   2013).   The   research   design   envisaged   a   set   of   macro   phases.   The   initial   phase   consisted   in   the   creation  of  a  case  study  repository  through  the  identification  and  prioritization  of  potential  sources   of  information,  an  open  invitation  for  proposal  of  cases  through  web2.0  channels,  followed  by  the   definition   of   the   1st-­‐round   criteria   for   selecting   at   least   twenty   practices   and   the   information-­‐ oriented  selection  of  the  corresponding  case  studies  on  applications  of  ICT  solutions  for  governance   and  policy  modelling.  In  the  second  phase,  case  studies  have  been  elicited  through  the  definition  of   the  2nd-­‐round  criteria  for  selecting  eight  promising  practices  and  the  application  of  a  multi-­‐criteria   method,   followed   by   further   elaboration   on   the   eight   case   studies   that   have   been   selected   by   the   13  |  P a g e  
  • 14.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   multi-­‐criteria   method   based   on   desk   research.   In   the   third   phase   the   final   four   cases   have   been   selected   and   subjected   to   an   in-­‐depth   analysis   carried   out   through   meticulous   study   of   the   available   public   documentation   and   the   conduction   of   interviews   with   key   involved   stakeholders.   After   the   final   selection   of   cases   and   the   in   depth   analysis,   the   findings   have   been   synthesized   through   the   analysis   of   the   emerging   trends   from   applications   of   ICT   solutions   for   governance   and   policy   modelling   as   well   as   the   development   of   key   considerations   for   the   CROSSOVER   roadmap   for   the   themes   that   refer   to   its   scope.   Finally   the   key   findings   of   the   analysis   of   the   four   cases   have   been   shared  with  the  CROSSOVER  partners  and  the  community  that  follows  closely  the  Policy  Making  2.0   domain   over   various   Web   2.0   channels,   to   provide   feedback   and   validation.   The   key   results   of   the   case  studies  are  described  later  in  the  impact  section.     7) Analysis  of  the  Prize   This   prize   was   given   to   the   best   policy-­‐making   2.0   applications,   that   is   are   for   the   best   use   of   technology   to   improve   the   design,   delivery   and   evaluation   of   Government   policy.   The   focus   of   the   jury  has  been  on  implementations  that  can  show  a  real  impact  on  policy  making,  either  in  terms  of   better  policy  or  wider  participation.  These  technologies  included,  but  are  not  limited  to:   • Visual  analytics   • Open  and  big  data   • Modelling  and  simulation  (beyond  general  equilibrium  models)   • Collaborative  governance  and  crowdsourcing   • Serious  gaming   • Opinion  mining   An   important   condition   for   participating   to   the   selection   has   been   the   real-­‐life   implementation   of   technology  to  policy  issues.     Out  of  50  applications,  the  jury  selected  the  best  12  and  eventually  the  3  winners,  which  received  an   IPAD  mini.    The  principal  domains  of  the  applications  were  as  follow:   • • • • • • 23  in  the  “Collaborative  Governance  and  Crowd-­‐sourcing”  domain   13  in  the  “Open  and  Big  Data”  domain   4  in  the  “Visual  Analytics”  domain   2  in  the  “Modelling  and  Simulation  (beyond  general  equilibrium  models)”  domain   2  in  the  “Serious  Gaming”  domain   1   in   each   of   the   following   domains:   “Open   Source   Governance”,   “Opinion   Mining”,   “Participatory  Policy  Making”     All  the  relevant  applications  received  have  been  integrated  in  the  roadmap.  The  criteria  for  judging   the  applications  were:   • • • • Impact  on  the  quality  of  policies   Openness,  scalability  and  replicability   Extensiveness  of  public  and  policymakers’  take  up   Technological  innovativeness   To  this  respect,  the  applicants  to  the  prize  were  required  to  provide  the  following  information:   • Name  of  the  application     14  |  P a g e  
  • 15.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • • • • • • • Year  of  launch     Short  description  of  the  technological  domain   Link  to  the  application     Describe  the  impact  of  the  application  on  the  quality  of  policies     Describe  the  public  and  policymaker  take  up  of  the  application   Describe  to  what  extent  the  application  was  technologically  innovative   Contact  details  of  the  applicant       8) LinkedIn  Group  Policy-­‐Making  2.0     A   crucial   element   in   the   engagement   of   stakeholders   is   given   by   the   creation   of   a   group   on   LinkedIn   called   Policy   Making   2.0 4 ,   which   is   a   virtual   place   where   actual   and   potential   practitioners   of   advanced  ICT  tools  for  policy-­‐making  can  exchange  experiences.  The  group  displays  a  high  selected   pool   of   high   level   members   (over   840)   engaging   in   discussions   and   exchange   of   views.   In   order   to   foster  debate  in  the  group,  the  CROSSOVER  consortium  posts  on  a  regular  base  info  about  the  new   cases   and   tools   to   be   integrated   in   the   knowledge   repository.   Some   other   discussion   topics   relate   to   the  best  ways  to  engage  the  government  in  online  policy  making,  the  posting  of  third  parties  content   and   info   about   incoming   CROSSOVER   workshops.   In   particular   the   group   is   being   used   for   disseminating  the  Survey  on  the  ICT  Needs  of  Policy  Makers,  as  well  as  the  roadmap  in  commentable   format.   The   Policy   Making   2.0   group   also   serves   as   a   liaison   channel   with   similar   projects   such   as   eGvoPoliNet   and   OCOPOMO.   As   agreed   the   eGovPoliNet   LinkedIn   group   has   merged   with   the   CROSSOVER   Policy   Making   2.0   group,   and   after   the   end   of   the   CROSSOVER   project   the   interaction   will   continue   led   by   the   eGovPoliNet   consortium.   Moreover   as   we   are   approaching   the   end   of   the   project  we  decided  to  shift  from  a  closed  LinkedIn  group  to  an  open  one.                                                                                                                               4   15  |  P a g e  
  • 16.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP         1.3. Scope  and  definition   Policy-­‐making  2.0  refers  to  a  set  of  methodologies  and  technological  solutions  aimed  at  innovating   policy-­‐making.  As  we  will  describe  in  section  2.1,  the  scope  goes  well  beyond  the  focus  on  “Decision-­‐ making”  notion  typical  of  eParticipation,  and  encompasses  all  phases  of  the  policy  cycle.  The  main   goal   is   limited   to   improving   the   quality   of   policies,   not   of   making   them   more   consensual   or   representative.   Policy-­‐making  2.0  is  a  new  term  that  we  have  coined  to  express  in  more  understandable  terms  the   somehow  technical  notion  of  “ICT  for  governance  and  policy  modelling”.  Its  usage  in  the  course  of   the  project  proved  more  effective  than  the  latter  when  discussing  with  stakeholders.  Thereby  from   now  on  we  will  refer  to  the  roadmap  as  the  Research  Roadmap  on  Policy-­‐Making  2.0.   The  full  set  of  methodologies  and  tools  has  been  spelled  out  in  the  taxonomy  in  WP15:   1.1.   Open  government  information  &  intelligence  for  transparency   1.1.1.   Open  &  Transparent  Information  Management  Open  data  policy  Open  data  licence  Open  data  portal  Code  list  Vocabulary/ontology  Reference  data  Data  cleaning  and  reconciliation  tool   1.1.2.   Data  published  on  the  Web  under  an  open  licence  Human-­‐readable  data  Machine  readable  data  in  proprietary  format  Machine-­‐readable  data  published  in  a  non-­‐proprietary  format  Data  published  in  RDF  SPARQL  endpoint  for  querying  RDF  data  RDF  data  linked  to  other  data  sets   1.1.3.   Visual  Analytics  Visualisation  of  a  single,  static,  embedded  data  set  Visualisation  of  multiple  static  data  sets  Visualisation  of  a  single  live  data  feed  or  updating  data  set  Visualisation  of  multiple  data  points,  including  live  feeds  or  updates   1.2.   Social  computing,  citizen  engagement  and  inclusion   1.2.1.   Social  Computing  Collaborative  writing  and  annotation  Content  syndication  Feedback  and  reputation  management  systems  Social  Network  Analysis  Participatory  sensing   1.2.2.   Citizen  Engagement                                                                                                                             5  The  taxonomy  presented  here  builds  on  CROSSROAD  taxonomy,  which  has  been  expanded,  reviewed  and  updated  by  the   members  of  the  Consortium   16  |  P a g e  
  • 17.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   1.3.   1.4.  Online  deliberation  Argumentation  support  Petition,  Polling  and  voting  Serious  games  Opinion  mining   1.2.3.   Public  Opinion-­‐Mining  &  Sentiment  Analysis  Opinion  tracking  Multi-­‐lingual  and  Multi-­‐Cultural  opinion  extraction  and  filtering  Real-­‐time  opinion  visualisation  Collective  Wisdom  Analysis  and  Exploitation   Policy  Assessment   1.3.1.   Policy  Context  Analysis  Forecasting  Foresight  Back-­‐Casting  Now-­‐Casting  Early  Warning  Systems  Technology  Road-­‐Mapping  (TRM)   1.3.2.   Policy  Modelling  Group  Model  Building  Systems  Thinking  &  Behavioural  Modelling  System  Dynamics  Agent-­‐Based  Modelling  Stochastic  Modelling  Cellular  Automata   1.3.3.   Policy  Simulation  Multi-­‐level  &  micro-­‐simulation  models  Discrete  Event  Simulation  Autonomous  Agents,  ABM  Simulation,  Multi-­‐Agent  Systems  (MAS)  Virtual  Worlds,  Virtual  Reality  &  Gaming  Simulation  Model  Integration  Model  Calibration  &  Validation   1.3.4.   Policy  Evaluation  Impact  Assessment  Scenarios  Model  Quality  Evaluation  Multi-­‐Criteria  Decision  Analysis   Identity,  privacy  and  trust  in  governance   1.4.1.   Identity  Management  Federated  Identity  Management  Systems  User  centric,  self  managed  and  lightweight  credentials  Legal-­‐social  aspects  of  eIdentity  management  Mobile  Identity  (Portability)   1.4.2.   Privacy  Privacy  and  Data  Protection  Privacy  Enhancing  Technologies  Anonymity  and  Pseudonymity  Open   data   management   (including   Citizen   Profiling,   'digital   shadow'   tracing   and  tracking   1.4.3.   Trust   17  |  P a g e  
  • 18.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   1.5.  Legal  Informatics  Digital  Rights  Management  Digital  Citizenship  Rights  and  feedback  loops  Intellectual  Property  in  the  digital  era  Trust-­‐building   Services   (including   data   processing   and   profiling   by   private   actors  for  public  services)   Future  internet  for  collaborative  governance   1.5.1.   Cloud  Computing  Cloud  service  level  requirements  Business  models  in  the  cloud  Cloud  interoperability  Security  and  authentication  in  the  cloud  Data  confidentiality  and  auditability  Cloud  legal  implications   1.5.2.   Pervasive  Computing  &  Internet  of  Things  in  Public  Services  Ambient  intelligence  Exploiting  smart  objects  Standardization  Business  models  for  pervasive  technologies  Privacy  implications  and  risks   1.5.3.   Provision  of  next  generation  public  e-­‐services  Fixed  and  mobile  network  access  technologies  Mobile  web  Models  for  information  dissemination  Management  of  scarce  network  capacity  and  congestion  problems  Large-­‐scale  resource  sharing  Interworking  of  different  technologies  for  seamless  connectivity  of  users   1.5.4.   Future  Human/Computer  Interaction  Applications  &  Systems  Web  accessibility  User-­‐centered  design  Augmented  cognition  Human  senses  recognition     Policy-­‐making  2.0  encompasses  clearly  a  wide  set  of  methodologies  and  tools.  At  first  sight,  it  might   appear   unclear   what   the   common   denominator   is.   In   our   view,   what   they   share   is   that   they   are   designed  to  use  technology  in  order  to  inform  the  formulation  of  more  effective  public  policies.  In   particular,   these   technologies   share   a   common   approach   in   taking   into   account   and   dealing   with   the   full   complexity   of   human   nature.   As   spelled   out   originally   in   the   CROSSOVER   project   proposal:   “traditional   policy-­‐making   tools   are   limited   insofar   they   assume   an   abstract   and   unrealistic   human   being:  rational  (utility  maximizing),  consistent  (not  heterogeneous),  atomised  (not  connected),  wise   (thinking   long-­‐term)   and   politically   committed   (as   Lisa   Simpson)”.   Policy-­‐making   2.0   thus   accounts   for   this   diversity.   Its   methodologies   and   tools   are   designed   not   to   impose   change   and   artificial   structures,   rather   to   interact   with   this   diversity.   Agent-­‐based   models   account   for   the   interaction   between   agents   that   are   different   in   nature   and   values;   systems   thinking   accounts   for   long-­‐term   interacting   impacts;   social   network   analysis   deals   with   the   mutual   influences   between   people   rather   than   fully   rational   choices;   big   data   analyses   observed   behaviour   rather   than   theoretical   models;   persuasive   technologies   deal   with   the   complex   psychology   of   individuals   and   introduces   gaming   values   to   involve   more   “casual”   participants.   Moreover,   policy-­‐making   2.0   tools   allow   all   stakeholders  to  participate  to  the  decision-­‐making  process.   18  |  P a g e  
  • 19.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     1.4. Policy:  Between  politics  and  services   The  application  of  technology  to  governmental  issues  is  not  a  new  topic.  Indeed  e-­‐government  and   the   new   buzzword   of   government   2.0,   have   become   mainstream   in   recent   years:   how   and   why   a   future  looking  research  agenda  could  still  refer  to  the  2.0  paradigm  as  innovative?  The  novelty  lies  in   the  “policy”  part  of  the  definition.   So  far,  the  application  of  "2.0"  technologies  to  governmental  processes  has  focussed  mainly  on  the   usage   of   social   media   for   political   communication,   best   exemplified   by   the   Obama   campaign.   The   typical  narrative  is  that  in  the  age  of  social  media,  traditional  communication  campaigns  and  political   parties   are   unsuited   to   generate   commitment   and   action   by   citizens,   which   instead   want   to   take   active   part   in   the   campaign   and   self-­‐organize   via   social   media:   ""A   candidate   who   can   master   the   Internet  will  not  only  level  the  playing  field;  he  will  level  the  opposition."  RightClick  Strategies'  Larry   Purpuro.   A  second  area  of  strong  focus  proved  to  be  the  collaborative  provision  of  public   services  based  on   peer-­‐to-­‐peer   support   and   open   data,   best   exemplified   by   the   widely   spread   "appsfordemocracy"   contests.   The   narrative   here   is   that   government   should   act   as   a   platform   and   enable   third   parties   (and  citizens  themselves)  to  co-­‐create  and  deliver  public  services  based  on  open  government  data.     This  is  what  Goldsmith  and  Eggers  (2004)  call  "governing  by  network".   Indeed,   the   Obama   administration   clearly   shows   these   priorities,   moving   from   state-­‐of-­‐the-­‐art   campaigning   in   order   to   be   elected,   and   then   implementing   a   strong   open   data   policy   with   crowdsourcing  initiatives  to  let  citizens  create  services  based  on  these  data.   Between   "politics"   and   "public   services   co-­‐delivery",   much   less   attention   has   been   devoted   to   the   usage  of  social  technology  to  improve  public  policy.  While  politics  deal  with  the  legislative  branch,   the   Parliament,   policy-­‐making   is   mainly   the   realm   of   the   executive   branch.   Typically,   the   job   of   policy-­‐making   involves   a   great   deal   of   socio-­‐economic   analysis   as   well   as   consultation   with   stakeholders.     This  roadmap  aims  to  fill  this  gap,  by  providing  a  complete  picture  of  how  technology  can  improve   policy-­‐making.     19  |  P a g e  
  • 20.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   2. Not  just  another  hype:  the  Demand  side  of  policy-­‐making  2.0   In  the  context  of  new  technologies,  we  are  periodically  informed  about  the  emerging  wave  that  will   change  everything,  only  to  see  it  quickly  forgotten  after  years  or  even  month  in  what  Gartner  calls   “trough  of  disillusionment”.  While  some  of  this  emphasis  is  certainly  driven  by  commercial  interests,   in  many  other  cases  it  reflects  a  genuine  optimism  of  its  proponents,  who  tend  to  underestimate  the   real-­‐life  bottlenecks  to  adoption  by  less  enthusiast  people.     Movzorov   critically   calls   this   cyber-­‐utopianism   or   technological   solutionism   (Morozov   2013);   on   a   similar   note,   many   years   of   eGovernment   policy   have   revealed   the   fundamental   importance   of   non-­‐ technological  factors,  such  as  organisational  change,  skills,  incentives  and  culture.     One   way   to   prevent   policy-­‐making   2.0   to   become   yet   another   hype   in   the   Gartner   curve,   is   to   precisely   spell   out   the   challenges   that   these   new   technologies   help   to   address.   Indeed,   the   importance  of  this  demand-­‐driven  approach  based  on  grand  challenges  is  fully  embraced  by  the  new   Horizon2020   research   programme   of   the   European   Union. 6     Furthermore,   a   demand-­‐driven   approach  helps  us  to  frame  the  technological  opportunities  in  a  language  understandable  to  policy-­‐ makers,  thereby  supporting  the  awareness-­‐raising  objective  of  the  CROSSOVER  project.   When   analysing   the   demand   side,   our   first   consideration   is   that   policy-­‐making   is   more   important   and   complex   than   ever.     The   role   of   government   has   substantially   changed   over   the   last   twenty   years.  Governments  have  to  re-­‐design  their  role  in  areas  where  they  were  directly  involved  in  service   provision,   such   as   utilities   but   also   education   and   health.   This   is   not   simply   a   matter   of   privatisation,   or   of   a   linear   trend   towards   smaller   government.   Indeed,   even   before   the   recent   financial   turmoil   and  nationalisation  of  parts  of  the  financial  system,  government  role  in  the  European  societies  was   not   simply   “diminishing”,   but   rather   being   transformed.   At   the   same   time,   it   is   increasingly   recognized  that  the  emergence  of  new  and  complex  problems  requires  government  to  increasingly   collaborate   with   non-­‐governmental   actors   in   the   understanding   and   in   the   addressing   of   these   challenges7.  As  an  OECD  report  states  the  following:     “Government   has   a   larger   role   in   the   OECD   countries   than   two   decades   ago.   But   the   nature   of   public   policy  problems  and  the  methods  to  deal  with  them  are  still  undergoing  deep  change.  Governments   are   moving   away   from   the   direct   provision   of   services   towards   a   greater   role   for   private   and   non-­‐ profit  entities  and  increased  regulation  of  markets.  Government  regulatory  reach  is  also  extending  in   new   socio-­‐economic   areas.   This   expansion   of   regulation   reflects   the   increasing   complexity   of   societies.   At   the   same   time,   through   technological   advances,   government’s   ability   to   accumulate   information   in   these   areas   has   increased   significantly.   As   government   face   more   new   and   complex   problems  that  cannot  be  dealt  with  easily  by  direct  public  service  provision,  more  ambitious  policies   require  more  complex  interventions  and  collaboration  with  non-­‐governmental  parties”   This  is  particularly  challenging  in  our  "complex"  societies.  “Complex”  systems  are  those  where  “the   behaviour  of  the  system  as  a  whole  cannot  be  determined  by  partitioning  it  and  understanding  the   behaviour   of   each   of   the   parts   separately,   which   is   the   classic   strategy   of  the  reductionist  physical   sciences”.  The  present  challenges  governments  must  face,  as  described  by  the  OECD,  are  complex  as   they   are   characterised   by   many   non-­‐linear   interactions   between   agents;   they   emerge   from   these   interactions   and   are   therefore   difficult   to   predict.   The   financial   crisis   is   probably   the   foremost   example   of   a   complex   problem,   which   proved   impossible   to   predict   with   traditional   decision-­‐making   tools.                                                                                                                             6     7  See  Ostrom:­‐lecture.html   20  |  P a g e  
  • 21.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     2.1. The  typical  tasks  of  policy-­‐makers:  the  policy  cycle   Policy-­‐making  is  typically  carried  out  through  a  set  of  activities  described  as  "policy-­‐cycle"  (Howard     2005).   In   this   document   we   propose   a   new   way   of   implementing   policies,   by   first   assessing   their   impacts  in  a  virtual  environment.   While   different   versions   of   the   cycle   are   proposed   in   literature,   in   this   context   we   adopt   a   simple   version  articulated  in  5  phases:   -­‐ agenda  setting  encompasses  the  basic  analysis  on  the  nature  and  size  of  problems  at  stakes   are  addressed,  including  the  causal  relationships  between  the  different  factors   -­‐ policy   design   includes   the   development   of   the   possible   solutions,   the   analysis   of   the   potential  impact  of  these  solutions8,  the  development  and  revision  of  a  policy  proposal   -­‐ adoption   is   the   cut-­‐off   decision   on   the   policy.   This   is   the   most   delicate   and   sensitive   area,   where  accountability  and  representativeness  are  needed.  It  is  also  the  area  most  covered  by   existing  research  on  e-­‐democracy     -­‐ implementation  is  often  considered  the  most  challenging  phase,  as  it  needs  to  translate  the   policy   objectives   in   concrete   activities,   that   have   to   deal   with   the   complexity   of   the   real   world  .  It  includes  ensuring  a  broader  understanding,  the  change  of  behaviour  and  the  active   collaboration  of  all  stakeholders.   -­‐ Monitoring   and   evaluation   make   use   of   implementation   data   to   assess   whether   the   policy   is   being  implemented  as  planned,  and  is  achieving  the  expected  objectives.   The   figure   below   (authors’   elaboration   based   on   Howard   2005   and   EC   2009)   illustrates   the   main   phases  of  the  policy  cycle  (in  the  internal  circle)  and  the  typical  concrete  activities  (external  circle)   that  accompany  this  cycle.  In  particular,  the  identified  activities  are  based  on  the  Impact  Assessment   Guidelines  of  the  European  Commission  (EC  2009).                                                                                                                             8  A   very   important   element   in   policy   design   and   formulation   is   given   by   ex-­‐ante   evaluation.   In   this   respect   ICT   tools   for   policy-­‐making  can  play  an  important  role,  simulating  alternative  policy  options  and  impacts  before  implementing  a  policy   action   21  |  P a g e  
  • 22.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     Figure  3:  Policy  Cycle  and  Related  Activities       Traditionally,  the  focus  about  the  impact  of  technology  in  policy-­‐making  has  been  on  the  adoption   phase,   analysing   the   implications   of   ICT   for   direct   democracy.   In   the   context   of   the   CROSSOVER   project,  we  adopt  a  broader  conceptual  framework  that  embraces  all  phases  of  policy-­‐making.     2.2. The  traditional  tools  of  policy-­‐making   Let   us   present   now   what   are   the   methodologies   and   tools   already   traditionally   adopted   in   policy-­‐ making.   Typically,   in   the   agenda-­‐setting   phase,   statistics   are   analysed   by   government   and   experts   contracted  by  government  in  order  to  understand  the  problems  at  stake  and  the  underlying  causes   of  the  problems.  Survey  and  consultations,  including  online  ones,  are  frequently  used  to  assess  the   stakeholders’  priorities,  and  typically  analysed  in-­‐house.  General-­‐equilibrium  models  are  used  as  an   assessment  framework.   Once  the  problems  and  its  causes  are  defined,  the  policy  design  phase  is  typically  articulated  through   an  ex-­‐ante  impact  assessment  approach.  A  limited  set  of  policy  options  are  formulated  in  house  with   22  |  P a g e  
  • 23.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   the   involvement   of   experts   and   stakeholders.   For   each   option,   models   are   simulated   in   order   to   forecast  possible  sectoral  and  cross-­‐sectoral  impacts.  These  simulations  are  typically  carried  out  by   general-­‐equilibrium   models   if   the   time   frame   is   focused   on   short   and   medium   term   economic   impacts  of  policy  implementation.  Based  on  the  simulated  impact,  the  best  option  is  submitted  for   adoption.   The   adoption   phase   is   typically   carried   out   by   the   official   authority,   either   legislative   or   executive   (depending   on   the   type   of   policy).   In   some   cases,   decision   is   left   to   citizens   through   direct   democracy,   through   a   referendum   or   tools   such   as   participatory   budgeting;   or   to   stakeholders   through  self-­‐regulation.   The   implementation   phase   typically   is   carried   out   directly   by   government,   using   incentives   and   coercion.   It   benefits   from   technology   mainly   in   terms   of   monitoring   and   surveillance,   in   order   to   manage  incentives  and  coercion,  for  example  through  the  database  used  for  social  security  or  taxes   revenues.   The  monitoring  and  evaluation  phase  is  supported  by  mathematical  simulation  studies  and  analysis   of   government   data,   typically   carried   out   in-­‐house   or   by   contractors.   Moreover,   as   numbers   aggregate   the   impacts   of   everything   that   happens,   including   policy,   it   is   difficult   to   single   out   the   impacts   of   one   policy   ex   post.   Final   results   are   published   in   report   format,   and   fed   back   to   the   agenda  setting  phase.     2.3. The  key  challenges  of  policy-­‐makers   Needless  to  say,  the  current  policy-­‐making  process  is  seldom  based  on  objective  evidence  and  not  all   views   are   necessarily   represented.   Dramatic   crises   seem   to   happen   too   often,   and   governments   struggle  to  anticipate  and  deal  with  them,  as  the  financial  crisis  has  shown.  Citizens  feel  a  sense  of   mistrust  towards  government,  as  shown  by  the  decrease  in  voters  turnout  in  the  elections.   In  this  section,  we  analyse  and  identify  the  specific  challenges  of  policy-­‐making.  The  goal  is  to  clearly   spell  out  "what  is  the  problem"  in  the  policy  making  process   that  policy-­‐making  2.0  tools  can  help  to   solve.   The  challenges  have  been  identified  on  desk-­‐based  research  of  "government  failure"  in  a  variety  of   contexts,  and  are  illustrated  by  real-­‐life  examples.   One   first   overarching   challenge   is   the   emergence   of   a   distributed   governance   model.   The   traditional  division  of  “market”  and  “state”  no  longer  fits  a  reality  where  public  decision  and  action  is   effectively  carried  out  by  a  plurality  of  actors.  Traditionally,  the  policy  cycle  is  designed  as  a  set  of   activities  belonging  to  government,  from  the  agenda  setting  to  the  delivery  and  evaluation.  However   in  recent  years  it  has  been  increasingly  recognized  that  public  governance  involves  a  wide  range  of   stakeholders,  who  are  increasingly  involved  not  only  in  agenda-­‐setting  but  in  designing  the  policies,   adopting   them   (through   the   increasing   role   of   self-­‐regulation),   implementing   them   (through   collaboration,  voluntary  action,  corporate  social  responsibility),  and  evaluating  them  (such  as  in  the   case   of   civil   society   as   watchdog   of   government).   As   Elinor   Ostrom   stated   in   her   lecture   delivered   when  receiving  the  Nobel  Prize  in  Economics9:  “A  core  goal  of  public  policy  should  be  to  facilitate  the   development   of   institutions   that   bring   out   the   best   in   humans.   We   need   to   ask   how   diverse   polycentric   institutions   help   or   hinder   the   innovativeness,   learning,   adapting,   trustworthiness,   levels   of   cooperation   of   participants,   and   the   achievement   of   more   effective,   equitable,   and   sustainable   outcomes   at   multiple   scales”.   This   acknowledgement   leads   to   important   implications   for   the                                                                                                                             9­‐lecture.html   23  |  P a g e  
  • 24.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   CROSSOVER   roadmap:   policy-­‐making   2.0   tools   are   not   just   tools   for   government,   but   for   all   stakeholders  to  participate  in  the  policy-­‐making  process10.     2.3.1. Detect  and  understand  problems  before  they  become  unsolvable   The  continuous   struggle   for   evidence-­‐based  policy-­‐making  can  have  some  important  and  potentially   negative   implications   in   terms   of   the   capacity   of   prompt   identification   of   problems.   Policy-­‐makers   have   to   balance   the   need   for   prompt   reaction   with   the   need   for   justified   action,   by   distinguishing   signal  from  noise.  Delayed  actions  are  often  ineffective;  at  the  same  time,  short-­‐term  evidence  can   lead   to   opposite   effects.   In   any   case,   government   have   scarce   resources   and   need   to   prioritize   interventions  on  the  most  important  problems.   For  instance  the  significant  underestimation  of  the  risks  of  the  housing  bubble  in  the  late  2000s,  and   the  systemic  reaction  that  it  would  lead  to,  led  to  delayed  reactions11.     Systemic  changes  do  not  happen  gradually,  but  become  visible  only  when  it's  too  late  to  intervene  or   the   cost   of   intervening   is   too   high.   For   example,   ICT   is   today   recognized   as   a   key   driver   of   productivity  and  growth,  but  evidence  to  prove  this  became  available  at  a  distance  of  years  from  the   initial   investment.     In   fact   the   initial   lack   of   correlation   between   ICT   investment   and   productivity   growth   was   mostly   due   to   incorrect   measurement   of   ICT   capital   prices   and   quality.   Subsequent   methodologies   found   that   computer   hardware   played   an   increasing   role   as   a   source   of   economic   growth   (see   inter   al.   Colecchia   and   Schreyer   2002,   Jorgenson   and   Stiroh   2000,   Oliner   and   Sichel   2000).   The  problem  is  in  this  case  is  therefore  twofold:  to  collect  data  more  rapidly;  and  to  analyse  them   with   a   wider   variety   of   models   that   account   for   systemic,   long   term   effects   and   that   are   able   to   detect  and  anticipate  weak  signals  or  unexpected  wild  cards.   2.3.2. Generate  high  involvement  of  citizens  in  policy-­‐making   The   involvement   of   citizens   in   policy-­‐making   remains   too   often   associated   with   short-­‐termism   and   populism.     It   is   difficult   to   engage   citizens   in   policy   discussions   in   the   first   place:   public   policy   issues   are   not   generally  appealing  and  interesting  as  citizens  fail  to  understand  the  relevance  of  the  issues  and  to   see   "what's   in   it   for   me".   The   decline   in   voters’   turnout   and   the   lack   of   trust   in   politicians   reflects   this.   More   importantly,   there   are   innumerable   cases   where   the   "right"   policies   are   not   adopted   because  citizens  "would  not  understand"  or  because  it  is  not  politically  acceptable.   While   the   Internet   has   long   promised   an   opportunity   for   widespread   involvement,   e-­‐participation   initiatives   often   struggle   to   generate   participation.   Participation   is   often   limited   to   those   that   are   already  interested  in  politics,  rather  than  involving  those  that  are  not.   When   participation   occurs,   online   debates   tend   to   focus   on   eye-­‐catching   issues   and   polarized   positions,  in  part  because  of  the  limits  of  the  technology  available.  It  is  extremely  difficult  and  time   consuming  to  generate  open,  large  scale  and  meaningful  discussion.       2.3.3. Identify  “good  ideas”  and  innovative  solutions  to  long-­‐standing  problems                                                                                                                             10  However   in   our   project   we   mainly   focus   on   tools   that   are   used   or   can   be   adopted   by   Governments,   otherwise   we   would   risk  to  enlarge  too  much  the  scope  of  the  research  roadmap   11­‐j26.html   24  |  P a g e  
  • 25.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   Innovation  in  policy-­‐making  is  a  slow  process.  Because  of  the  technical  nature  of  issues  at  hand,  the   policy   discussion   is   often   limited   to   restricted   circles.   Innovative   policies   tend   to   be   "imported"   through   "institutional   isomorphism".   Innovative   ideas,   from   both   civil   servants   and   citizens,   fail   to   surface  to  the  top  hierarchy  and  are  often  blocked  for  institutional  resistance.   Existing   instruments   for   large-­‐scale   brainstorming   remain   limited   in   usage,   and   fail   to   surface   the   most   innovative   ideas.   Crowdsourcing   typically   focus   on   the   most   “attractive”   ideas,   rather   than   the   most  insightful.   2.3.4. Reduce  uncertainty  on  the  possible  impacts  of  policies     When  policy  options  have  been  developed,  simulations  are  carried  out  to  anticipate  the  likely  impact   of   policies.   The   option   with   the   most   positive   impact   is   normally   the   one   that   is   proposed   for   adoption.   Most  existing  methodologies  and  tools  for  the  simulation  of  policy  impacts  work  decently  with  well   known,  linear  phenomena.  However,  they  are  not  effective  in  times  of  crisis  and  fast  change,  which   unfortunately  turn  out  to  be  exactly  the  situations  where  government  intervention  is  most  needed.     As  an  example  nowadays  the  European  Central  Bank  bases  its  analysis  of  the  EURO  Area  economy   and   monetary   policy   on   a   derived   version   of   the   DSGE   model   developed   by   Frank   Smet   and   Raf   Wouters   in   2003 12 .   Smet   and   Wouters’   model   is   deeply   microfounded,   allowing   for   a   rigorous   theoretical   structure   of   the   model.   Moreover   in   this   setting   the   reduced   form   parameters   are   related  to  deep  structural  parameters  in  order  to  mitigate  Lucas’  critique,  while  the  utility  of  agents   can  be  taken  as  a  measure  of  welfare  in  the  economics  (Phelps  ed.  1970).     However,   the   DSGE   models   suffer   from   several   shortcomings   jeopardizing   their   ability   to   predict,   let   alone  to  prevent,  a  global  crisis:   • Agents   are   assumed   to   be   perfectly   rational,   having   perfect   access   to   information   and   adapting  instantly  to  new  situations  in  order  to  maximize  their  long-­‐run  personal  advantage   • So   far   agents   have   entered   the   models   as   homogeneous   representative   entities,   while   it   would  be  a  step  forward  being  able  to  take  into  account  agents  heterogeneity   • Canonical  models  consider  atomistic  agents  with  little  or  no  interactions  and  thereby  are  not   able  to  cope  with  network  externalities     But  most  of  all  it  is  the  very  notion  of  equilibrium  which  prevents  standard  models  from  dealing  with   crisis.   A   stable   steady   state   equilibrium   is   a   condition   according   to   which   the   behaviour   of   a   dynamical   system   does   not   change   over   time   or   in   which   a   change   in   one   direction   is   a   mere   temporary   deviation.   This   condition   is   proper   of   general   equilibrium   theory,   in   which   a   stable   steady   state   is   believed   to   be   the   norm   rather   than   the   exception.   When   in   the   canonical   model   we   are   out   of  equilibrium,  the  situation  is  seen  just  as  a  short  lapse  before  the  return  to  the  steady  state.  This  is   in   sharp   contrast   with   the   very   notion   of   crisis,   which   represents   a   steady   deviation   from   the   equilibrium.  Loosely  speaking,  the  crisis  phenomenon  is  not  even  conceived  within  the  framework  of   standard  models.   All  these  flaws  are  not  only  related  to  DSGE  models,  but  also  to  Computational  General  Equilibrium   (CGE)   or   macro-­‐econometric   forecasting   models,   which   are   the   traditional   policy   making   tools.   In   this   view   it   would   be   very   important   to   find   new   frameworks   capable   of   avoiding   those   shortcuts.                                                                                                                             12   25  |  P a g e  
  • 26.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   Some   of   such   methodologies   and   methods   already   exists   and   some   governments   are   using   them.   Our  aim  is  to  push  forward  in  that  direction.   We   need   to   move   away   from   the   equilibrium   paradigm   in   order   to   be   able   to   assess   other   issues:   evolutionary   dynamics;   heterogeneity   of   technologies   and   firm;   political   and   legal   determinants   of   social   stability;   incentive   structures;   better   modelling   technological   change,   innovation   diffusion   and   economic   systems   (taking   into   account   finance,   debt   and   insurance);   interactions   between   heterogeneous   economic   agents   (firms   and   households)   and   central   governments;   heterogeneous   responses  to  government  incentives;  economic  dependence  from  the  ecosystem.       Trichet,   the   former   head   of   ECB,   clearly   put   it:   “This   doesn't   mean   we   have   to   abandon   DSGE...(but)...atomistic   rational   agents   don't   capture   behaviour   during   a   crisis...rational   expectations  theory  has  brought  macroeconomics  a  long  way  ...  but  there  is  a  clear  case  to  re-­‐ examine  the  assumptions”   But  the  need  for  new  policy  making  tools  is  not  limited  to  the  economic  realm:  in  the  future  it  will   become   more   and   more   important   to   anticipate   non-­‐linear   potentially   catastrophic   impacts   from   phenomena  such  as:  climate  change  (draught  and  global  warming);  threshold  climate  effects  such  as   poles’  sea-­‐ice  withdraw,  out-­‐gassing  from  melting  permafrost,  Indian  monsoon,  oceans  acidification;   social   instability   affecting   economic   well-­‐being   (social   conflict,   anarchy   and   mass   people   movements).       The   lack   of   understanding   of   systemic   impact   has   driven   to   short   term   policies   which   failed   in   grasping  long  term,  systemic  consequences  and  side  effects:   -­‐ An   example   of   this   approach   might   be   given   by   the   sovereign   debt   issue.   In   fact   it   is   relatively  easy  for  governments  under  popular  pressure  to  increase  expenditure  and  public   debt   to   cope   with   short   term   necessities,   such   as   offsetting   the   negative   impacts   brought   about  by  a  regional  or  global  crisis.  On  the  other  hand  it  is  harder  to  take  into  account  the   long   term   effect   determined   by   higher   interest   rates   on   private   investments   and   consumption  through  crowding  out  and  fiscal  pressure.   -­‐ Another  example  of  short-­‐termism  are  the  financial  policies  pursued  in  south  East  Asia  at  the   beginning   of   the   90s.   Many   countries,   such   as   Thailand,   liberalized   their   financial   markets   fostering  the  inflow  of  investments  aimed  at  sustaining  growth.  Unfortunately  those  capitals   triggered  a  real  estate  bubble  which  has  been  at  the  roots  of  the  1997-­‐1998  crisis.   -­‐ In  2008  the  Central  Bank  of  Iceland  yielded  liquidity  loans  for  saving  banks  on  the  verge  of   default  on  the  basis  of  newly-­‐issued,  uncovered  bonds,  i.e.  effectively  printing  fiat  money  on   demand,   causing   a   significant   rise   in   inflation.   To   cope   with   this   rise   in   prices,   the   Iceland   Central  Bank  had  to  keep  very  high  interest  rates  thereby  leading  to  an  economic  bubble.   -­‐ According   to   a   large   number   of   economists   the   financial   crisis   was   triggered   by   US   government   policies   spanning   across   two   administrations   which   were   intended   to   ensure   citizens’   right   but   instead   determined   an   unprecedented   high   number   of   risky   mortgages,   as   well   as   the   decline   in   mortgage   underwriting   standards   that   ensued.   According   to   the   “Financial  Crisis  Inquiry  Commission  Report”  13  those  policies,  together  with  the  deregulation   of  the  financial  system,  might  have  catalysed  the  crisis.                                                                                                                             13   26  |  P a g e  
  • 27.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   -­‐ Other   examples   can   be   the   bail   out   of   financial   institutions:   in   the   short   run   those   actions   maintain   employment   and   economic   standards,   while   in   the   long   term   they   induce   moral   hazard,  keep  operating  inefficient  companies  and  decrease  the  trust  of  economic  agents  in   regulation,  which  is  the  funding  pillar  of  our  economic  system.   2.3.5. Ensure  long  -­‐  term  thinking   In   traditional   economics,   decisions   are   utility-­‐maximising.   Agents   rationally   evaluate   the   consequences  of  their  actions,  and  take  the  decision  that  maximize  their  utility.  However,  it  is  well   known   that   this   rationalistic   view   does   not   fully   capture   human   nature.   We   tend   to   overestimate   short-­‐term  impact  and  underestimate  the  long  term.  In  policy-­‐making,  short-­‐termism  is  a  frequent   issue.   People   are   reluctant   to   accept   short-­‐term   sacrifices   for   long-­‐term   benefits.   Politicians   have   elections   typically   every   5   years,   and   often   their   decisions   are   taken   to   maximize   the   impact   “before   the   elections”.   There   is   also   the   perception   that   laypeople   are   less   sensitive   to   long   term   consequences,   which   are   instead   better   understood   by   experts.   Overall,   long-­‐term   impact   is   less   visible  and  easier  to  hide,  due  to  lack  of  evidence  and  data.  As  a  result,  decisions  are  too  often  taken   looking  at  short-­‐term  benefits,  even  though  they  might  bring  long  term  problems.     Climate  change  is  a  typical  policy  area  where  sub-­‐optimal  decisions  were  taken  because  the  short-­‐ term  costs  were  considered  to  outweigh  the  long  term  consequences.  The  long  term  impact  is  not   visible,   while   the   short   term   sacrifices   were,   even   though   ICT   had   an   important   role   in   stimulating   the  debate  and  catalysing  attention  of  the  media  on  the  issue.     2.3.6. Encourage  behavioural  change  and  uptake   Once   policies   are   adopted,   a   key   challenge   is   to   make   sure   that   all   stakeholders   comply   with   regulations  or  follow  the  recommendations.  It  is  well  known  how  the  greatest  resistance  to  a  policy   is  not  active  opposition,  but  lack  of  application.   For  instance,  several  programmes  to  reduce  alcohol  dependency  problems  in  the  UK  failed  as  they   excessively   relied   on   positive   and   negative   incentives   such   as   prohibition   and   taxes,   but   did   not   take   into  account  peer-­‐pressure  and  social  relationships.  They  failed  to  leverage  “the  power  of  networks”   (Ormerod   2010).   For   instance,   any   policy   related   to   reduction   of   alcohol   consumption   through   prohibitions  and  taxes  is  designed  to  fail  as  long  as  it  does  not  take  into  account  social  networks,  as   binge   drinkers   typically   have   friends   who   also   have   similar   problems.   In   another   classical   example   (Christakis  and  Fowler  1997),  a  large  scale  longitudinal  study  showed  that  the  chances  of  a  person   becoming  obese  rose  by  57  per  cent  if  he  or  she  had  a  friend  who  became  obese.   The   identification   of   social   networks   and   the   role   of   peer   pressure   in   changing   behaviour   is   not   considered  in  traditional  policy-­‐making  tools.   2.3.7. Manage  crisis  and  the  “unknown  unknown”   The   job   of   policy-­‐makers   is   increasingly   one   of   crisis   management.   There   is   robust   evidence   that   the   world   is   increasingly   interconnected,   and   unstable   (also   because   of   climate   change).   Crises   are   by   definition   sudden   and   unpredictable.   Dealing   with   unpredictability   is   therefore   a   key   requirement   of   policy-­‐making,  but  the  present  capacity  to  deal  with  crises  is  designed  for  a  world  where  crises  are   exceptional,  rather  than  the  rule.  Donald  Rumsfeld,  former  secretary  of  state,  famously  said  during   the  Iraq  war  that  while  the  US  government  was  capable  of  dealing  with  the  “known  unknown”,  the   difficulty  was  the  increasing  recurrence  of  “unknown  unknown”:  those  things  that  we  don’t  known   that  we  don’t  know.   There   is   evidence   that   the   instability   and   chaotic   natures   of   our   world   is   increasing,   because   of   its   increasing   connectedness.   Every   year,   intense   climate   phenomena   throw   our   cities   in   disarray,   27  |  P a g e  
  • 28.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   because   of   snow,   flooding,   fires.   Each   crisis   seems   to   find   our   decision-­‐makers   unprepared   and   unable  to  deal  with  it  promptly.   As  Taleb  (2007)  puts  it,  we  live  in  the  age  of  "Extremistan":  a  world   of   "tipping   points"   (Schelling   1969)   “cascades”   and   "power   laws"   (Barabasi   2003)   where   extreme   events   are   "the   new   normal".   There   are   many   indications   of   this   extreme   instability,   not   only   in   negative   episodes   such   as   the   financial   crisis   but   also   in   positive   development,   such   as   the   continuous   emergence   of   new   players   on   the   market   epitomised   by   Google.   The   random   vulnerability  of  today’s  world  is  well  illustrated  by  this  chart  from  the  EC  DG  RESEARCH.     Figure  4:  Total  Disasters  Reported   2.3.8. Moving  from  conversations  to  action   The  collaborative  action  of  people  is  able  to  achieve  seemingly  unachievable  goals:  experiences  such   as   ZooGalaxy   and   Wikipedia   show   that   mass   collaboration   can   help   achieve   disruptive   innovation.   Yet   too   often   web-­‐based   collaboration   is   confined   to   complaints   and   discussions,   rather   than   action.   As  one  blogger  put  it,  paraphrasing  Marx:  “Philosophers  have  only  interpreted  the  world:  the  point  is   to  complain  about  it”14.     For  example,  the  2012  Italian  elections  saw  an  explosion  of  activity  in  social  media  discussing  about   the  different  candidates.  This  energy  then  failed  to  translate  into  concrete  action  in  the  aftermath  of   the  elections.         2.3.9. Detect  non-­‐compliance  and  mis-­‐spending  through  better  transparency   In  times  of  crisis,  it  is  ever  more  important  for  governments  to  ensure  that  financial  resources  are   well   spent   and   policies   are   duly   implemented.   But   monitoring   is   a   cost   in   itself,   and   a   certain   margin   of  inefficiency  in  resources  deployment  is  somehow  “natural”.  Yet  the  cost  of  this  mismanagement  is   staggering:  for  instance,  in  2010,  7.7%  of  all  Structural  Funds  money  was  spent  in  error  or  against  EU   rules15.     OECD   estimates   place   the   cost   of   corruption   equals   5%   of   global   GDP16.   Thereby   it   would   be   crucially  important  to  be  able  to  avoid  the  mismanagement    with  anticipatory  corrective  actions.   28  |  P a g e  
  • 29.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   2.3.10. Understand  the  impact  of  policies   Measuring   the   impact   of   policies   remains   a   challenge.   Ideally,   policy-­‐makers   would   like   to   have   real-­‐ time   clear   evidence   on   the   direct   impact   of   their   choice.   Instead,   the   effects   of   a   policy   are   often   delayed  in  time;  the  ultimate  impact  is  affected  by  a  multitude  of  factors  in  addition  to  the  policy.   Timely  and  robust  evaluation  remains  an  unsolvable  puzzle.   This   is   particularly   true   for   research   and   innovation   policy,   where   the   results   from   investment   are   naturally  expected  at  years  of  distance.  As  Kuhlmann  and  Meyer-­‐Krahmer  (1994)  puts  it,  “the  results   of  evaluations  necessarily  arrive  too  late  to  be  incorporated  into  the  policy-­‐making  process”.     2.4. When  policy-­‐making  2.0  becomes  a  reality:  a  tentative  vision  for   2030   This   is   the   scenario   of   how   future   policy-­‐making   could   be   deployed   in   an   ideal   world,   if   all   the   opportunities   of   policy-­‐making   2.0   tools   were   taken.   It   aims   at   illustrating   how   these   technologies   and   methods   could   concretely   be   deployed   and   the   effect   they   would   have.   This   is   deliberately   a   normative  scenario,  describing  a  positive  and  concrete  future  at  a  very  high  level.   The   scenario   is   organised   alongside   the   typical   phases   of   the   policy-­‐making   cycle.   It   applies   to   a   hypothetical  new  privacy  directive  being  developed  in  2030.     2.4.1. Agenda  setting  phase:  recognizing  the  problem   Brussels,   2030.  The  EC  task  force  on  privacy  and  data  protection  is  alerted  by  a  number  of  events.   Their   yearly   report,   accompanied   by   the   publication   as   linked   open   data   in   February,   has   been   accessed  by  more  than  10.000  people  in  a  week.  Several  high  profile  online  blogs  have  published  the   geo-­‐visualized   mash   up   of   the   task   force   data   with   the   data   from   customer   complaints   about   broadband   slowness.   The   figures   speak   for   themselves:   the   complaints   from   customers,   collected   through   both   the   government   single   feedback   system   and   social   media,   about   privacy   infringements   and  identity  theft  mirror  exactly  the  broadband  disruption.  All  seem  to  point  to  some  kind  of  "data   theft"   at   the   infrastructural   level   of   the   Internet.   A   similar   analysis   on   open   linked   data   shows   an   abnormal  concentration  of  complaints  over  credit  card  fraud  from  users  of  a  limited  number  of  ISP   that   have   struggled   to   obtain   the   infrastructure   security   certification.   Anomalies   in   this   correlation   seem  to  weaken  the  case,  but  are  quickly  discovered  when  social  network  analysis  is  carried  out:  not   only   the   users   of   ISP   but   also   their   friends   and   contacts   are   most   likely   to   report   denounces   for   fraud.   While  some  years  ago  the  task  force  members  would  still  address  this  through  the  traditional  slow   policy  process,  only  to  realize  its  social  impact  after  mass  media  take  this  up,  today  a  quick  look  at   social   media   analytics   confirms   that   the   public   is   deeply   concerned.   Hashtags   like   #wherearemydata   are   drawing   thousands   of   comments.   The   task   force   obtains   real   time   report   on   sentiment   and   opinions   being   shared   publicly;   it   appears   clear   that   people   feel   unprotected   by   existing   instruments   and  regulation  and  voice  their  dissatisfaction  mainly  towards  the  Task  Force  itself.  In  particular,  the                                                                                                                                                                                                                                                                                                                                                                                             14  Quoted   in   Mick   Fealty,   The   wisdom   of   crowds,   The   Guardian   24   February   2007   15­‐names-­‐worst-­‐managers-­‐of-­‐eu-­‐money/72613.aspx   16 29  |  P a g e  
  • 30.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   reputation   report   quickly   identifies   a   limited   numbers   of   social   media   activists   that   show   high   influence  in  terms  of  shaping  the  public  opinion  on  the  matter,  as  their  message  is  quickly  spread.   Historical   text   analysis   of   social   media   allows   to   predict   that   users   that   complain   over   privacy   infringement  are  likely  to  dramatically  decrease  the  extent  to  which  they  share  information  and  data   on   the   web   over   the   following   weeks.   This   drastic   cuts   to   content   sharing   becomes   a   serious   liability   for  an  economy  which  is  now  built  on  the  assumption  that  people  naturally  share  and  collaborate  on   the   web.   Reduction   in   knowledge   sharing,   as   predicted   by   social   media   analytics,   could   lead   to   a   reduction  of  economic  activity  which  is  already  fragile.   An  in-­‐depth  investigation  discovers  a  hardware  hacking  group  that  has  targeted  a  selected  number   of   lowly   protected   broadband   providers   to   steal   data   directly   from   their   traffic.   The   policy   agenda   of   the   Task   Force   are   quickly   switched   to   develop   a   revision   of   regulation   in   order   to   better   link   the   broadband  regulation  with  data  protection.     An  open  collaboration  group  is  convened,  with  the  direct  involvement  of  users  previously  identified   as  "highly  influential"  through  social  media  analytics.  In  addition,  cross-­‐analysis  reputation  tools  are   used   to   identify   "experts"   on   joined-­‐up   policy   approaches   to   data   protection   and   broadband   infrastructure,   based   on   integrated   data   from   social   media   (e.g.   Klout   and   Linkedin)   and   scientific   impact   (e.g.   Altmetrics,   ISI   impact   factor).   This   group   is   called   to   provide   independent   fact   based   analysis  of  the  problem  based  on  best  available  data.     In   particular,   the   group   is   called   to   understand   and   model   the   causal   relationship   between   fear   of   privacy   infringements   and   reduction   in   knowledge   sharing;   and   between   this   reduction   and   economic   growth.   The   analysis   is   carried   out   through   a   combination   of   network   analysis,   system   dynamics   and   agent   based   modelling.   Their   report   simulates   several   possible   scenarios,   but   the   common   theme   is   that   a   reduction   in   sharing   activities   by   key   influencers   could   lead   to   a   major   economic  downturn,  as  non-­‐sharing  behavior  will  soon  spread  from  the  "geeks"  to  the  general  public   through  imitation  and  social  pressure.     The  report  is  published  for  public  review,  enabling  in-­‐line  comments  and  in-­‐depth  analysis  of  the  raw   data  and  models  behind  the  analysis.  It  brings  in  hundreds  of  comments.  Once  a  quick  text  mining   analysis  is  carried  out,  the  comments  seem  to  cluster  on  an  unjustified  assumption  in  one  scenario,   and   on   a   limited   set   of   issue   regarding   the   potential   negative   impact   on   net   neutrality   when   implementing  new  regulation.  Hence,  the  scenario  is  double-­‐checked  and  the  assumption  clarified,   and  net  neutrality  experts  are  brought  on  board  in  the  working  group.     2.4.2. Policy  design   Once  the  nature  and  size  of  the  problem  is  clarified,  the  working  group  is  called  to  design  possible   policy  measures.  A  crowdsourcing  exercise  is  launched,  where  anyone  can  submit  ideas  for  specific   amendments   to   the   present   regulation.   The   analysis   is   based   on   the   most   voted   suggestion,   but   these  turn  out  to  be  not  extremely  insightful.  A  reputation  management  system  is  integrated  with   the   exercise,   allowing   to   identify   original   ideas   and   insight   based   on   voting   “weighted”   based   on   expertise  in  the  field.  A  set  of  10  recommendations  are  presented  for  further  analysis  by  the  working   group.   The  working  group,  based  on  this  input,  formulate  3  policy  options  alongside  these  axis:   -­‐  continuing  with  the  current  data  protection  framework   -­‐   enhancing   it   with   greater   forms   of   self-­‐regulation;   increased   transparency,   easier   enforcement   and   greater  empowerment  of  users   -­‐  define  a  new,  stricter  data  protection  regulation   30  |  P a g e  
  • 31.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     The   three   options   are   then   run   through   the   large-­‐scale   simulation   engine,   which   combines   agent   based  modeling,  system  dynamics,  network  analysis  and  big  data  analytics.  This  allows  to  anticipate   the  unexpected  effect  of  a  new  stricter  regulation,  that  would  probably  induce  virtuous  broadband   providers   to   conform   to   the   minimum   requested   by   regulation,   while   private   companies   that   typically  choose  the  most  secure  providers  would  have  to  increase  their  expenditure  on  security,  in   particular   in   the   field   of   web   services,   which   is   already   weak   in   Europe   and   exposed   to   global   competition.   In   addition,   consumers   would   be   satisfied   with   a   perception   of   increased   security,   thereby  reducing  their  attention  on  own  control  over  data,  which  would  in  the  long  term  increase   the  security  risks  of  another  crisis.     All   the   models   underlying   the   simulation   are   open   to   public   review,   in   order   to   ensure   the   transparency   of   the   initiative.   Indeed,   the   first   version   of   the   ex-­‐ante   impact   assessment   showed   that   option   1   was   the   less   risky   and   more   beneficial,   but   a   little   known   researcher   from   Greece   quickly  pointed  out  to  a  banal  coding  mistake  in  the  database  used  to  compute  historical  series  of   privacy  infringements.   In  the  end,  option  2  is  chosen  as  the  most  effective.  The  amendment  to  the  regulation  are  quickly   rapidly  drafted,  publicly  reviewed  and  then  turned  into  law  by  the  European  Parliament.       2.4.3. Implementation   The  regulation  envisages  a  strong  role  for  the  public  in  both  enforcement  and  self-­‐regulation.  Each   local  branch  of  broadband  providers  have  to  publish  in  real  time    as  open  linked  data  the  results  of   the  security  certification,  as  well  as  any  traffic  management  intervention  they  carry  out.     In   addition,   a   set   of   “persuasive   games”   has   been   developed   to   help   consumers   manage   and   control   their  data  flows.  Users  receive  badges  each  time  they  perform  a  data  safety  self-­‐assessment,  which   is   easy   to   carry   out   through   a   highly   visual   smartphone   app   which   highlights   to   what   extend   the   users   behaviour   diverges   from   the   public   recommendations   and   from   the   people   in   his/her   social   network.   Unexpectedly,   a   kind   of   game   about   being   “safest   kid   on   the   block”   starts   particularly   between  teenagers,  that  compete  in  trying  to  overcome  each  other  safety  provisions.  New  business   models  of  third  party  data  management  services  are  launched  for  those  less  interested  in  managing   their  data.       As   a   result,   the   propensity   to   buy,   share   and   collaborate   online   increase   sensibly,   driving   to   a   moderately  positive  economic  impact.     2.4.4. Evaluation   Real-­‐time   data   analytics   on   the   performance   of   data   providers,   as   well   as   anonymized   data   on   consumers  data  protection  measure,  allow  decision-­‐makers  to  identify  potential  breaches  as  soon  as   they  happen.   Participatory  sensing  tools,  combined  with  opinion  mining  allow  citizens  and  policy-­‐makers  to  easily   monitor  when  new  problems  emerge.   Open  data  on  measures  taken  by  regulators  allow  civil  society  organisation  to  ensure  the  adequacy   of  government  intervention.     31  |  P a g e  
  • 32.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   2.5. The   key   challenges   for   policy   makers   and   the   corresponding   phases  in  the  policy  cycle   Let  us  now  relate  the  key  challenges  of  policy  making  activity  with  the  phases  in  the  policy  cycle:   • The  Agenda  Setting  phase  is  mostly  related  to  the  challenges   o Detect  and  understand  problems  before  they  become  unsolvable     o Manage  crisis  and  the  “unknown  unknown”     o Ensure  long  -­‐  term  thinking:  Agenda   • The  Design  phase  is  mostly  related  to  the  challenges     o Encourage  behavioural  change  and  uptake   o Identify  “good  ideas”  and  innovative  solutions  to  long-­‐standing  problems   o Reduce  uncertainty  on  the  possible  impacts  of  policies   o Generate  high  involvement  of  citizens  in  policy-­‐making   • The  Implementation  phase  is  mostly  related  to   o Moving  from  conversations  to  action   o Reduce  uncertainty  on  the  possible  impacts  of  policies   • The  Monitor  and  Evaluation  phase  is  mostly  related  to   o Detect  non-­‐compliance  and  mis-­‐spending  through  better  transparency     o Manage  crisis  and  the  “unknown  unknown”     32  |  P a g e  
  • 33.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   3. The  supply  side:  current  status  and  the  Research  Challenges     In  this  section,  we  illustrate  in  detail  each  research  challenge  which  needs  to  be  addressed  in  order   to  make  the  vision  a  reality  and  address  the  policy-­‐challenges  described  in  the  previous  chapter,  by   describing:   -­‐ The  definition,     -­‐ The  potential  opportunities  for  governance,     -­‐ The  state  of  the  art  of  market  and  research,     -­‐ The  existing  challenges  and     -­‐ The  recommended  research  themes.   The   research   challenges   are   organised   in   2   groups:   the   first   regroups   6   challenges   on   Policy   Modelling,  while  the  second  one  regroups  9  challenges  on  Collaborative  Governance.           3.1. Policy  Modelling     3.1.1. Systems  of  Atomized  Models     Introduction  and  definition   This   research   challenge   seeks   to   find   the   way   to   model   a   system   by   using   already   existing   models   or   composing   more   comprehensive   models   by   using   smaller   building   blocks,   sometimes   also   called   “atoms”,   either   by   reusing   existing   objects/models   or   by   generating/building   them   from   the   very   beginning.   Therefore,   the   most   important   issue   is   the   definition/identification   of   proper   (or   most   apt)  modelling  standards,  procedures  and  methodologies  by  using  existing  ones  or  by  defining  new   ones.   Further   to   that,   the   present   sub-­‐challenge   calls   for   establishing   the   formal   mechanisms   by   which  models  might  be  integrated  in  order  to  build  bigger  models  or  to  simply  exchange  data  and   valuable  information  between  the  models.  Finally,  the  issue  of  model  interoperability  as  well  as  the   availability   of   interoperable   modelling   environments   should   be   tackled,   as   well   as   the   need   for   feedback-­‐rich   models   that   are   transparent   and   easy   for   the   public   and   decision   makers   to   understand.     Why  it  matters  in  governance   Using   existing   objects/models   that   are   able   to   describe   systems,   sub-­‐systems   and   interaction   among   them,   allows   everyone   to   build   his   own   insight   on   a   specific   problem/solution.   So,   in   governance,   such  opportunity  gives  us  the  chance  to:   • Release   public   data,   linking   them   and   producing   visual   representations   able   to   reveal   unanticipated  insights.   33  |  P a g e  
  • 34.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Use  social  computing  to  promote  engagement  and  citizens’  inclusion  in  policy  decision,  and   exploit  the  power  of  ICT  in  mining  and  understanding  the  opinions  they  express.   • Analyse  policies  and  produce  models  that  can  be  visualised  and  run  to  produce  simulations   able  to  show  the  effects  and  impacts  from  different  perspectives  such  as  political,  economic,   social,  technological,  environmental  and  legal  facets.   Current  Practice  and  Inspiring  cases   In  systems  analysis,  it  is  common  to  deal  with  the  complexity  of  an  entire  system  by  considering  it  to   consist   of   interrelated   sub-­‐systems.   This   leads   naturally   to   consider   models   as   consisting   of   sub-­‐ models.   Such   a   (conceptual)   model   can   be   implemented   as   a   computer   model   that   consists   of   a   number   of   connected   component   models   (or   modules).   Component-­‐oriented   designs   actually   represent  a  natural  choice  for  building  scalable,  robust,  large-­‐scale  applications,  and  to  maximize  the   ease  of  maintenance  in  a  variety  of  domains.   An  implementation  based  on  component  models  has  at  least  two  major  advantages:     • First,   new   models   can   be   constructed   by   coupling   existing   component   models   of   known   and   guaranteed   quality   with   new   component   models.   This   has   the   potential   to   increase   the   speed  of  development.   • Secondly,  the  forecasting  capabilities  of  two  different  component  models  can  be  compared,   as  opposed  to  compare  whole  simulation  systems  as  the  only  option.       Further,   common   and   frequently   used   functionalities,   such   as   numerical   integration   services,   visualisation   and   statistical   ex-­‐post   analyses   tools,   can   be   implemented   as   generic   tools   and   developed   once   for   all   and   easily   shared   by   model   developers.   By   the   way,   the   current   practice   in   composing  and  re-­‐using  models  is  still  not  sufficiently  widespread.  In  relation  to  Model  Reuse,  this  is   mainly  due  to  the  fact  that  little  to  no  repository  actually  exists17.  Moreover,  the  publicly  available   models   are   not   “open”   to   modification   or   re-­‐use.   It   would   be   useful   if   every   paper   containing   a   model   included   a   link   to   on-­‐line   version   that   people   could   run   and   modify, Some   modelling   environments   (or   modelling   suites)   provide   some   examples   and   small   libraries   of   ready-­‐to-­‐use   models,   but   in   most   cases,   they   are   not   completely   open   nor   any   explanation   is   provided   on   how   to   reproduce  them  (their  structure,  parameters,  etc.).  As  an  inspiring  case  see  the  SEAMLESS  project,   which   was   funded   by   the   EU   Framework   Programme   6   (Global   Change   and   Ecosystems),   ran   from   2005   till   March   2009,   and   developed   a   computerized   framework   for   integrated   assessment   of   agricultural  systems  and  the  environment18.  During  the  project,  a  modular  approach  was  chosen  to   develop  a  system  named  “Agricultural  Production  and  Externalities  Simulator  (APES)”,  illustrated  in   figure  (5).  APES  is  a  modular  simulation  system  targeted  at  estimating  the  biophysical  behaviour  of   agricultural  production  systems  in  response  to  the  interaction  of  weather,  soils  and  different  options   of  agro-­‐technical  management.  Although  a  specific,  limited  set  of  components  is  available  in  the  first   release,   the   system   is   being   built   to   incorporate,   at   a   later   time,   other   modules   which   might   be   needed  to  simulate  processes  not  included  in  the  first  version.  The  processes  are  simulated  in  APES   with  deterministic  approaches  which  are  mostly  based  on  mechanistic  representations  of  biophysical   processes.   APES   was   used   to   compare   alternative   agricultural   and   environmental   policy   options,   facilitating  the  process  of  assessing  key  indicators  that  characterize  interactions  between  agricultural   systems,   natural   and   human   resources,   and   society.   The   developed   framework,   named   SEAMLESS-­‐IF                                                                                                                             17  This  is  true  for  most  of  the  sectors,  even  though  for  instance  most  energy  models  are  based  on  the  MARKAL   family  of  models.  Furthermore  something  to  consider  is  that  models  need  to  be  customized,  so  that  having  a   single  framework  readily  applicable  to  different  contexts  and  sectors  may  actually  be  counter  productive   18  http://www.seamless-­‐   34  |  P a g e  
  • 35.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   in   a   finale   stage,   also   enabled   linkage   of   quantitative   models,   pan-­‐European   databases   and   qualitative  procedures  to  simulate  the  impact  on  society  of  biophysical,  economic  and  behavioural   changes.   SEAMLESS-­‐IF   now   facilitates   ex-­‐ante   assessments   at   the   full   range   of   scales   from   the   global   to   the   field   level   to   support   policy   and   decision   making   for   sustainable   development.   SEAMLESS-­‐IF   nowadays   can   be   used   to   investigate   the   effects   of   agricultural   and   environmental   policies   while   accounting   for   technical   innovations.   Further,   the   interactions   of   such   policies   with   other   major   trends  such  as  climate  change  and  increasing  land  used  for  bio-­‐fuel  crops  can  be  studied  efficiently   in  the  near  future.   35  |  P a g e  
  • 36.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP                                                            Figure  5:  Agricultural  Production  and  Externalities  Simulator  (APES)       Analyses   with   SEAMLESS-­‐IF   can   be   done   at   multiple   scales   and   with   varying   time   horizons,   whilst   focusing  on  the  most  important  issues  emerging  at  each  scale.  This  is  possible  as  the  framework  is   based   on   research   innovations   in   linking   models   across   scales   allowing   consistent   “micro-­‐macro”   analysis   as   well   as   linking   models   across   disciplines   allowing   “economicbiophysical”   analysis.   The   linked   models   range   from   a   bio-­‐physical   field   model   to   a   farm   model   and   to   an   agricultural   sector   model  for  the  EU;  in  other  words  they  ensure  a  consistent  analysis  of  what  effects  EC  policies  may   have  on  agricultural  markets,  farming  systems  and  the  environment.  In  addition,  the  effectiveness  of   a   policy   in   its   institutional   context   is   assessed   by   applying   qualitative   procedures.   The   interlinked   pan-­‐European  database  provides  the  relevant  data  needed  at  different  scales.       For   another   inspiring   example   have   a   look   at   the   Insight   Maker   case   at   Insight  Maker  allows  to  build  simulation  models  ("Insights")  for  all  scales:  from  the  smallest  cell,  to   the  social  effects  of  product  adoption,  to  global  climate  change.  Once  they  are  built  one  can  share   them   with   others.   The   models   are   called   “an   Insight”   as   they   will   typically   reveal   one   or   more   fascinating  point  about  the  system  under  study.  All  the  simulations  built  with  Insight  Maker  can  be   shared  via  the  web.    This  means  people  can  change  the  variables  and  see  the  results  for  themselves.     Vensim  Molecules19  is  a  software  used  for  constructing  system  dynamics  models  from  molecules  of   system   dynamics   structure.   Molecules   are   made   of   primitive   stock   and   flow   or   auxiliary   elements   and   are,   in   turn,   the   building   blocks   of   complete   models,   elements   of   substructure   serving   a   particular   purpose.   Molecules   provide   a   framework   for   presenting   important   and   commonly   used   elements  of  model  structure  making  faster  and  easier  to  develop  system  dynamics  models.                                                                                                                             19   36  |  P a g e  
  • 37.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     Anylogic 20,   a   multi-­‐method   simulation   modelling   tool   capable   of   integrating   and   combining   the   following   modelling   approaches:   system   dynamics,   discrete   event   simulation   and   agent-­‐based   modelling.   Anylogic’s   simulation   language   is   composed   by   stock   and   flow   diagrams   (used   for   System   Dynamics   modelling),   statecharts,   which   define   the   agents’   behaviour   in   Agent   Based   modelling,   action   charts   (used   to   define   algorithms),   and   finally   process   flowcharts   which   are   the   basic   constructions  for  defining  processes  in  Discrete  Event  modelling.     Available  Tools   A   very   interesting   tool   is   En-­‐ROADS21,   a   global   simulation   model   that   focuses   on   how   changes   in   global   GDP,   energy   efficiency,   R&D   results,   carbon   price,   fuel   mix,   and   other   factors   will   change   carbon  emissions,  energy  access,  and  temperature.       En-­‐ROADS   is   designed   to   complement,   other   more   disaggregated   models   addressing   these   questions,  and  relies  on  the  other  models  and  EIA  projections  for  testing  and  data.                                                                                                                                 20   21­‐roads   37  |  P a g e  
  • 38.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP             En-­‐ROADS   is   customized   to   address   enquiries   regarding   how   much   might   technological   breakthroughs  contribute  to  addressing  climate  change.  These  particular  breakthroughs  include  for   instance  R&D  and  scale-­‐up  of  a  new  zero-­‐carbon  energy  supply,  renewable  energy,  energy  efficiency,   inexpensive   natural   gas,   etc.   More   precisely   En-­‐ROADS   investigates   which   assumptions   about   the   technology   and   the   economy   would   be   necessary   for   a   breakthrough   to   grow   with   enough   speed   and  scale  to  deliver  climate  goals.             38  |  P a g e  
  • 39.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     One   of   the   most   innovative   parts   of   En-­‐ROADS   regards   its   capabilities   to   test   assumptions   about   the   potential  success  of  R&D  towards  zero-­‐carbon  energy.  More  in  particular  the  simulation  investigate   what  are  the  likely  dynamics  of  the  emergence  of  a  new  energy  supply,  as  well  as  how  fast  could  it   grow  and  displace  high-­‐carbon  sources  and  reduce  carbon  emissions.  En-­‐ROADS  is  an  extension  of   the  C-­‐ROADS  model,  which  will  be  described  below  in  the  roadmap.  The  distinction  between  the  two   models  is  that  while  C-­‐ROADS  focuses  on  how  the  changes  in  national  and  regional  emissions  could   affect   GHG   emissions   and   climate   outcomes,   En-­‐ROADS   focuses   on   how   changes   in   the   energy,   economic,  and  public  policy  systems  could  influence  GHG  emissions  and  climate  outcomes.     Key  challenges  and  gaps   With  regards  to  implementation  architecture  and  use  of  modelling  frameworks,  there  are  two  major   problems:     • the   framework   design   and   implementation   must   be   optimized   to   balance   carefully   its   flexibility   and   its   usability   to   avoid   incurring   either   a   performance   penalty   or   users   having   too  steep  a  learning  curve,  and     • developing  components  for  a  specific  framework  constrains  their  use  to  that  framework.     The   most   immediate   option   to   overcome   such   problems   is   developing   inherently   reusable   components  (i.e.  non  framework  specific),  which  can  be  used  in  a  specific  modelling  framework  by   encapsulating  them  using  dedicated  classes  called  “wrappers”;  such  classes  act  as  bridges  between   the   framework   and   the   component   interface.   The   disadvantage   of   this   solution   is   the   creation   of   another   “layer”   in   the   implementation,   which   adds   to   the   already   implemented   machinery   in   the   framework.   The   appropriateness   of   this   solution,   both   as   ease   of   implementation   and   overall   performance,  must  be  evaluated  case  by  case.   Regardless   of   the   choice   of   developing   framework   specific   or   intrinsically   reusable   components,   there   is   a   basic   choice   which   must   be   carefully   evaluated   prior   to   that   and   which   is   related,   in   general   terms,   to   the   framework   as   a   flexible   modelling   environment   to   build   complex   models   39  |  P a g e  
  • 40.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   (model   linking),   but   also   to   the   framework   as   an   efficient   engine   for   simulation,   calibration   and   simulation   of   model   components   (model   execution).   Modern   software   technologies   allow   building   flexible,   coherent   and   elegant   constructs,   but   that   comes   at   a   performance   cost.   Without   even   introducing  specific  references  to  Object  Oriented  Programming  (OOP),  it  seems  important  to  point   out   that   the   use   of   object-­‐oriented   programming   constructs,   which   actually   enhance   flexibility,   modularity  and  reuse  of  software,  all  nice  things,  require  the  compiler  to  use  virtual  methods  calls,   dynamic   dispatching,   and   so   on.   All   these   operations   are   resource   intensive   and   in   some   cases,   they   can  heavily  affect  the  code  performance,  and  this  becomes  evident  in  applications  in  which  such  use   is  done  thousand  times  every  simulation  step.     By   the   way,   the   Model   Composition   horizon   is   even   more   clouded   as   the   potential   advantages   resulting  from  the  possibility  of  composing  bigger  models  from  smaller  ones  have  been  shown  only   recently.   It   is   essentially   due   to   the   problem   of   interoperability   and   integration   of   different   vendors’   (thus   proprietary)   model   formats   and   to   the   lack   of   standards   allowing   performing   composition   tasks.   Another   problem   stems   from   the   fact   that   many   models   are   still   too   dependent   on   their   implementation  methodology.  Moreover,  model  integration  is  at  present  almost  non-­‐existing.  Very   few   modelling   environments/suites   provide   the   import/export   functionalities   and   a   standard   language  for  model  interoperability  is  not  currently  available.  Most  of  the  current  practice  for  data   communication   or   information   transfer   is   performed   by   means   of   third   party   solutions   (e.g.:   interoperability   in   most   cases   is   achieved   by   transferring   data   via   electronic   spreadsheets   or,   only   in   rare  cases,  by  using  Database  Management  Systems  (DBMS)  or  Enterprise  Resource  Planning  (ERP)   systems.     Current  research   Current  research,  as  well  as  previous  research,  has  not  yet  worked  on  (with  the  exception  of  just  a   few  cases)  the  problem  of  different  models  integration.  At  present,  due  to  the  plethora  of  different   modelling/simulation  environments/suites,  as  well  as  to  differences  at  the  scientific  field  level,  many   competing   file   formats   exist.   It   is   possible   that   vendors   perceive   the   modelling   practice   as   a   very   small  market  niche  (as  the  users  stem  mainly  from  Academia  and  to  a  very  small  extent  from   private   companies  where  a  Decision  Support  Systems  is  used,  what  is  more  the  Public  Administration  share   is  negligible)  and  therefore  are  reluctant  to  introduce  interoperable  features.   Also,  current  research,  as  well  as  previous  research,  has  only  recently  begun  to  explore  the  following   issues:   • Open-­‐source  modelling  and  simulation  environments  (there  are  open  environments  that  are   rising  in  importance  in  the  research  community,  albeit  in  most  cases  they  only  provide  the   possibility  to  implement  and  simulate  a  model  according  to  the  modelling  methodology  they   refer  to).   • Communication   of   data   among   models   developed   in   different   proprietary   (or   open)   environments   by   depending   on   third   party   solutions   (e.g.:   interoperability   is   in   most   cases   only   achieved   by   transferring   data   by   means   of   electronic   spreadsheets   or,   only   in   rare   cases,  by  using  a  DBMS  or  an  organisation’s  ERP).   • Open   visualisation   of   results   stemming   from   model   simulation   (e.g.:   online   visualisation   of   simulation  results  in  a  browser  by  interfacing  -­‐  only  in  a  few  cases  -­‐  the  simulation  engines,   or  -­‐  as  it  is  more  often  the  case  -­‐  by  connecting  to  a  third  party  mean,  as  described  in  the   previous  bullet  point).   40  |  P a g e  
  • 41.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     Future  research   Future  research  should  therefore  focus  on:   • Definition   of   standard   procedures   for   model   composition/decomposition,   e.g.   how   to   deductively   pass   from   a   macro-­‐description   of   models   to   the   fine   definition   of   its   building-­‐ blocks   or   molecules   (top-­‐down   approach),   how   to   inductively   conceive   a   progressive   composition   of   bigger   models   by   aggregating   new   modules   as   soon   as   they   are   needed   (bottom-­‐up  approach)  or  by  expanding  already  existing  objects.   • Proposition   of   a   minimum   set   of   archetypical   structures,   building   blocks   or   molecules   that   might   be   used   according   to   the   proper   level   of   decomposition   of   the   model   (e.g.   systemic   archetypes,  according  to  the  Systems  Thinking  /  System  Dynamics  approach,  might  be  useful   to   describe   the   overall   behaviour   thanks   to   the   main   variables   in   the   system   to   be   modelled   at   a   macro-­‐to-­‐middle   level).   The   procedures   to   implement,   validate   and   redistribute   any   further  improvement  of  these  “minimal”  objects  should  be  investigated.   • Definition   of   open   modelling   standards,   as   the   basis   for   interoperability,   that   is   defining   common  file  formats  and  templates  (i.e.:  by  means  of  XML),  which  would  allow  the  models   described   by   means   of   these   XML   files   to   be   opened,   accessed   and   integrated   into   every   (compliant)  model-­‐design  and  simulation  environment22.   • Interoperability,  also  intended  in  terms  of  Service  Oriented  Architectures  (e.g.:  certain  stand-­‐ alone  and  always  operative  models  might  expose  some  “services”  in  order  to  make  available   either  their  endogenous  data  or  bits  of  information,  or  some  peculiar  function  or  structural   part,  while  some  other  may  request  to  use  those  services  when  needed.  In  consequence,  it   creates   a   need   for   a   definition   of   model   repositories,   a   list   of   operative   models   and   the   functionalities   that   they   might   expose   which   finally,   entails   the   definition   of   a   SOA   among   interoperable  models).   • Definition   and   implementation   of   model   repositories   (and   procedures   to   add   new   objects   to   them),   even   if   they   are   restricted   to   hosting   models   developed   according   to   a   specific   methodology  (Agent  Based,  System  Dynamics,  Event  Oriented,  Stochastic,  etc.)   • Definition  and  implementation  of  new  relationships  that  are  created  when  two  models  are   integrated.   All   possible   important   relationships   resulting   from   a   model   integration/composition   should   be   identified   and   eventually   included   in   the   new   deriving   integrated  model.   • Input   /   Output   definition   /   re-­‐definition:   the   integration   of   modelling   techniques   is   a   pertinent   issue   in   the   scope   of   this   challenge.   The   multi-­‐modelling   tools   should   be,   in   the   future,  available  not  only  to  experts  but  also  to  lay  users.  Moreover,  at  present,  only  a  few  of   the  actually  available  modelling/simulation  suites  are  able  to  provide  the  possibility  to  build   a  model  by  referring  to  a  different  modelling  methodology.                                                                                                                                 22     Making   portability   is   very   hard   to   implement:   in   the   past   there   has   been   a   significant   effort   with   SMILE   and   later   XMILE   to   make   SD   models   portable   between   different   software   programs.   This   was   really   a   best   case   scenario   as   the   software   programs   in   play   were   so   very   similar   to   start,   but   it   has   ended   up   unsuccessfully   as   agree   and   implementation   has   not   been   reached.   Deciding   on   a   universal   format   to   bind   existing   things   together   seems   like   it   will   not   be   able   to   work   (  There  has  been  greater  success  with  things  like  Modelica  where  you  get  a  format  and  an  app  and   then  other  programs  decide  to  adopt  the  format  themselves,  a  more  organic  process   41  |  P a g e  
  • 42.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   3.1.2. Collaborative  Modelling     Introduction  and  definition   The  English  sayings  “two  heads  are  better  than  one”  and  “too  many  cooks  spoil  the  broth”  give  an   idea   of   the   expectations   that   arise   from   a   collaboration   of   people.   On   the   one   hand,   one   would   expect  that  a  group  of  people  is  able  to  better  observe  and  perceive  situations  as  well  as  to  make   better  decisions  than  a  single  person  would  be  able  to  make.  On  the  other  hand,  it  is  also  common   knowledge   that   the   collaboration   of   several   people   entails   the   problem   of   group   coordination,   which,   if   disregarded,   can   make   group   work   inefficient,   compared   to   the   work   of   a   single   person.   There  is  the  need  for  an  authoritative  gatekeeper  that  is  also  the  modeler  implementing  the  model,   otherwise   collaborative   sessions   would   get   out   of   control   quickly   with   non-­‐implementable   ideas   becoming  the  focus  of  discussion  very  fast.     There  are  three  kinds  of  problems  that  are  typically  approached  by  groups:     • cognition  problems,  problems  with  a  definite  solution  or  a  set  of  solutions  that  are  certainly   better  than  others;   • coordination   problems,  problems  that  require  the  group  to  figure  out  how  to  coordinate  the   behaviour  of  its  members;   • cooperation   problems,   problems   which   feature   the   involvement   of   several   self-­‐interested,   distrustful  people  who  have  to  work  together.       Collaborative   modelling   (also   called   group   model   building)   refers   to   a   process   where   a   number   of   people  actively  contribute  to  the  creation  of  a  model.  The  weakest  form  of  involvement  is  feedback   to   the   session   facilitator,   similar   to   the   conventional   way   of   modelling.   Stronger   forms   are   proposals   for  changes  or  (partial)  model  proposals.  In  this  particular  approach  the  modelling  process  should  be   supported   by   a   combination   of   narrative   scenarios,   modelling   rules,   and   e-­‐Participation   tools   (all   integrated  via  an  ICT  e-­‐Governance  platform):  so  the  policy  model  for  a  given  domain  can  be  created   iteratively   using   cooperation   of   several   stakeholder   groups   (decision   makers,   analysts,   companies,   civic  society,  and  the  general  public).     As  a  matter  of  fact  groups  require  rules  (or  cultural  norms)  to  maintain  order  and  coherence,  as  well   as   diversity   and   independence   of   its   group   members   in   order   to   create   a   kind   of   a   collective   intelligence.   Bringing   together   people   with   diverse   perspectives   and   backgrounds   for   working   together   in   multi-­‐disciplinary   teams   is   expected   to   improve   the   overall   group   performance,   so   the   first  issue  on  which  the  collaborative  process  should  be  based  is  the  definition  of  a  shared  modelling   rules  framework  (the  social  norms),  guiding  the  modelling  team  in  determining  whether  a  proposal   is  accepted  or  rejected.  Two  usually  adopted  types  of  rules  are:   • Rules   of   majority,   where   a   certain   number   of   group   members   had   to   support   or   oppose   a   proposal  in  order  for  the  whole  group  to  accept  or  reject  it  (e.g.,  more  than  half).  A  tie-­‐break   rule   was   sometimes   specified   (e.g.,   for   the   case   of   an   equal   number   of   supporters   and   opponents).  The  tie-­‐break  could  involve  seniority  issues.   • Rules   of   seniority,  where  the  weight  of  a   group   member’s   support   or   opposition   was   related   to  his  or  her  status  within  the  group.  This  status  could  be  acquired  (e.g.,  by  experience)  or   associated  with  a  position  to  which  the  member  was  appointed.  A  frequent  example  of  this   was   the   case   of   a   more   experienced   modeller   who   was   considered   as   the   leader   by   the   group  and  took  decisions  on  their  behalf.  The  other  members  filled  the  role  of  consultants  in   such  a  case.   42  |  P a g e  
  • 43.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     These  rules  were  sometimes  set  up  explicitly  before  the  group  began  their  work,  or  in  an  early  phase   of   this   work.   But   in   most   cases   they   rather   emerged   as   the   result   of   each   member’s   behaviour.   Individuals   making   regular   contributions   of   high   quality   were   likely   to   acquire   seniority.   In   homogeneous  teams  majority  rules  were  used  more  often.     Why  it  matters  in  governance   From   a   very   high   level   of   abstraction,   collaborative   modelling   itself   can   be   seen   as   a   social   interaction   between   several   people,   while   these   people   who   together   perform   the   modelling   process  form  a  social  entity.  Thus,  the  process  of  collaboratively  defining  and  implementing  a  model,   with  a  particular  reference  to  the  public  policy  modelling,  is  strictly  connected  with  the  public  aspect   of   every   citizen’s   life,   starting   from   the   communities   bridged   by   the   decision   makers   that   collaboratively  define  some  policies,  to  an  average  citizen  which  interacts  with  other  citizens  within   the  rules  framework  defined  by  the  policies  themselves.     Starting   from   the   needs   perceived   by   the   citizens,   the   limitations   of   existing   modelling   techniques   adopted  in  policy  making  include  the  following  issues:   • Changing   models   is   too   time-­‐consuming   and   integrating   to   other   diagrams   is   difficult.   Also   there  are  version  control  problems.     • It  is  not  possible  for  more  than  one  person  to  work  on  the  same  diagram  at  the  same  time.     • Modelling  has  to  be  done  at  the  specific  location  where  the  modeller  is  present.   • Contribution  to  the  model  comes  from  those  interviewed  or  at  a  group  meeting,  limiting  the   potential  contribution  from  a  larger  group.   • Low  model  acceptance:  the  model  resulting  from  the  modelling  session  is  not  supported  by   some  of  the  stakeholders.   • Participants   feel   misunderstood:   as   a   consequence   of   bad   elicitation   or   a   wrong   understanding  of  the  model.   • Low   perceived   model   quality   and   limited   model   comprehension:   Individuals   do   not   fully   understand  the  model  or  do  not  agree  with  it.     Reasons  that  argue  for  conducting  policy  modelling  in  a  collaborative  manner  are:   • No   person   typically   understands   all   requirements   and   understanding   tends   to   be   distributed   across  a  number  of  individuals.     • A  group  is  better  capable  of  pointing  out  shortcomings  than  an  individual.     • Individuals   who   participate   during   analysis   and   design   are   more   likely   to   cooperate   during   implementation.   43  |  P a g e  
  • 44.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     Collaborative   modelling   calls   for   the   definition   of   the   citizen’s   role   in   the   public   policy   modelling   process   (e.g.:   the   mass   participation   issues   and   processes   have   been   already   researched   in   depth   by   the   e-­‐Participation   research   programs).   In   order   to   guarantee   participation   there   are   some   prerequisites  that  should  be  fulfilled:   • All   citizens   who   access   ICT   services   in   order   to   participate   should   represent   the   views   of   communities  affected  by  the  given  policy;   • All   citizens   are   able   to   take   part   in   the   modelling   process   via   intuitive   IT   systems   that   enable   them  an  effective  and  efficient  contribution;   • All   citizens   possess   proper   skills   (or   are   assisted)   to   purposely   follow   a   process   of   group   model-­‐building   in   order   to   avoid/abate   wrong   mental   models   and   thus   ultimately   reach   a   shared  vision  of  the  problem.                                                                                                                                                                                                                                                                             Current  Practice  and  Inspiring  cases   In   current   practice,   collaborative   modelling   is   mainly   performed   offline;   still   the   rules   and   guidelines   for   session   processes   are   not   yet   sufficiently   widespread.   In   fact,   the   abatement   of   wrong   mental   models   and   the   creation   of   knowledge   from   information   usually   imply   the   dialogue   among   people   with   different   views   of   the   problem   as   well   as   the   need   for   critical   skills.   Further   to   that,   the   information   that   occurred   in   a   discussion   has   to   be   grounded   and   definitively   transferred   to   the   formal   model.   Thus,   e-­‐Participation   might   be   of   help   in   achieving   a   critical   mass   of   data   and   information   exchange   online   but   in   itself   does   not   solve   the   problem   of   mass   cooperation   and   collaboration  in  a  formal  modelling  process.  Even  more,  the  participation  in  this  process  entails,  at   present,   a   thorough   knowledge   on   modelling   processes   or   tools   that   an   average   citizen   does   not   have.   Therefore,   there   is   an   urgent   need   for   Intuitive   Interfaces,   Modelling   Wizards   and   guided   simplified  approaches  to  modelling.  Starting  from  the  relevance  of  collaborative  modelling  in  policy   making,   as   a   very   former   inspiring   case   Maarten   Sierhuis   and   Albert   M.   Selvin,   working   at   NYNEX   Science  &  Technology  Inc  in  New  York,  presented  in  1996  a  applied  research  report  on  “Towards  a   Framework   for   Collaborative   Modelling   and   Simulation”,   describing   methodologies   for   modelling   and   simulation   in   a   collaborative   analysis   or   design   project,   and   describing   a   case   study   in   which   Conversational   Modelling,   a   software-­‐supported   technique   for   collaborative   modelling,     enabled   participants  to  construct  static  knowledge  models  in  collaborative  sessions.  The  sessions  described   in   the   report   resulted   in   the   identification   of   207   queries.   Of   these,   24   were   chosen   for   detailed   modelling.   As   a   result   of   the   modelling,   44   resources,   29   knowledge   items,   58   data   items,   and   8   organizational  issues  were  identified.  The   response   from   participants   was   positive.   Many   stated   that   they   had   learned   more   about   each   other   work   in   the   conversational   modelling   sessions   then   they   had   been   able   to   in   the   course   of   their   normal   work   activities.   The   development   organization   has   been   able   to   use   the   output   of   the   sessions   to   generate   design   requirements.   A   picture   of   the   interface  (figure  6)  used  during  the  sessions  follows.   44  |  P a g e  
  • 45.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP          Figure  6:  Conversational  Modelling  Interface       As   more   recent   inspiring   case,   one   can   refer   to   the   results   of   the   FP7   projects   OCPOMO 23  or   PADGETS24.   This   last   one,   PADGETS,   aims   at   bringing   together   two   well   established   domains,   the   mashup   architectural   approach   of   web   2.0   for   creating   web   applications   (gadgets)   and   the   methodology   of   system   dynamics   in   analysing   complex   system   behaviour.   The   objective   is   to   design,   develop   and   deploy   a   prototype   toolset   that   will   allow   policy   makers   to   graphically   create   web   applications   that   will   be   deployed   in   the   environment   of   underlying   knowledge   in   Web   2.0   media.   The  project  introduces  the  concept  of  Policy  Gadget  (PADGET)  –  similarly  to  the  approach  of  gadget   applications   in   web   2.0   –   to   represent   a   micro   web   application   that   combines   a   policy   message   with   underlying  group  knowledge  in  social  media  (in  the  form  of  content  and  user  activities)  and  interacts   with  end  users  in  popular  locations  (such  as  social  networks,  blogs,  forums,  news  sites,  etc.)  in  order   to  get  and  convey  their  input  to  policy  makers.                                                                                                                               23    Open  Collaboration  in  Policy  Modelling,   24   45  |  P a g e  
  • 46.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP                        Figure  7:  the  PADGET  Framework     PADGET  is  composed  of  four  main  components:   • A  message,  that  is  a  policy  in  any  of  its  stages  and  forms   • A   set   of   interaction   services,   that   allows   users   to   interact   with   the   policy   gadget   (find   it,   access  its  content,  comment  its  content,  share  it  etc.).  These  interfaces  may  be  provided  by   either  the  underlying  social  media  platforms  in  which  the  PADGET  Campaign  is  launched  or   by   the   PADGET   itself   when   it   takes   the   form   of   a   micro   application   (i.e.   in   the   case   of   the   iGoogle  gadget).   • The  social  context,  that  is  the  framework  describing  the  social  activity  and  content  relating   with   the   policy   gadget   in   each   individual   social   media   platform   where   the   policy   gadget   is   present.   • The   decision   services,   which   are   offered   by   two   modules.   The   PADGETS   analytics   and   the   PADGETS   simulation   model.   The   decision   services   component   is   responsible   for   the   generation   of   the   information   outputs   to   be   presented   to   the   PADGET   initiator   (usually   a   policy  maker).       PADGETS  will  use  publicly  available  APIs  for  interconnecting,  publishing  and  retrieving  content  from   underlying  social  media  platforms.  The  collected  information  and  user  activities  that  policy  gadgets   invoke   in   the   media   platforms   will   be   categorized   using   semantic   tags   as   to   their   relation   to   the   policies  in  order  to  help  the  policy  maker  form  an  opinion  about  what  the  users  think  about  relevant   issues  and  policies.       46  |  P a g e  
  • 47.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   An   interesting   model   is   Threshold   21   (T21),   built   using   the   System   Dynamics   methodology   to   facilitate  decision  making.  Feedback  relations  among  key  variables  within  sectors  are  endogenously   simulated  by  T21,  and  those  relations  can  lead  to  further  feedback  to  other  sectors.  This  process  is   valuable  for  learning  more  about  the  complex  interactions  among  and  across  sectors  that  need  to  be   taken   into   account   in   order   to   develop   more   effective   policies   and   mitigate   or   avoid   negative   side   effects.  This  approach  allows  to  bring  experts  together  from  different  sectors  to  better  understand   these  relations  and  obtain  the  data  needed  to  translate  a  qualitative  causal  diagram  (a  map  of  the   system)  into  a  quantitative  model,  using  a  participatory  approach.       Causal Diagram Example Society Economy + government expenditure income per capita R + Health-GDP loop + - + + health R + productivity Nutrition loop nutrition + + + + agriculture production + population GDP - - + B employment Agriculture -water loop + water stress B + + Health-GDP -energy loop + water demand fuel prices + renewable energy - Legend: (+) Positive link emissions + + fossil fuel consumption Environment + (-) Negative link   Source:   “Macro   Economic   Policy   Analysis   Applications”,   presented   by   John   Shilling   at   the   Transatlantic  Research  on  Policy  Modelling  Workshop25       An   interesting   case   of   application   of   the   model   to   inform   policy   making   regarded   China.   More   in   particular,  a  number  of  agencies  and  NGOs  working  in  China  cooperated  with  the  MI  in  an  attempt   to   address   a   wide   variety   of   issues   related   to   achieving   more   sustainable   growth,   dealing   with   resource  constraints  (e.g.  water,  agriculture,  energy)  in  the  face  of  a  large  and  growing  population,   reducing  GHG  emissions,  and  promoting  more  innovation  to  move  down  a  greener  path.    The  effects   of   the   growth   in   population,   the   economy,   transportation,   energy   consumption,   food   imports   on   growth   prospects   have   been   examined.     The   study   identified   also   some   important   cross   sector   factors,   such   as   that   slowing   the   growth   of   the   per   capita   size   of   housing   units   would   have   many   beneficial   effects,   including   less   cement   and   steel   production,   leading   to   less   GHG   emissions,   less                                                                                                                             25  http://www.CROSSOVER-­‐   47  |  P a g e  
  • 48.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   energy  demand,  less  reduction  in  arable  land  for  agriculture.  These  illustrate  important  cross  sector   effects  and  show  how  policies  need  to  take  a  broad  view  of  their  results.    By  generating  scenarios   over  the  longer  term  till  2030,  the  model  was  able  to  show  how  continuing  business  as  usual  would   lead   to   significant   challenges   and   tipping   points   on   water   demand,   agricultural   production,   and   GHG   emissions.     Then   it   was   demonstrated   how   policies   that   would   shift   level   of   consumption   and   innovation  would  have  significant  impacts  on  sustainability,  including  that  some  slower  increase  in   overall   consumption   may   be   critical   for   achieving   sustainability   on   these   indicators,   despite   lower   GDP   growth   and   job   creation.     It   shows   that   it   is   important   to   develop   policies   that   mitigate   the   weaker   performance   while   assuring   sustainability   and   that   provide   everyone   with   an   acceptable   living   standard.     It   clearly   illustrates   the   policy   challenges   faced   and   lays   the   basis   for   developing   more  effective  policies.   Scenario summary for 2030 Low Consump Unit High Tech Baseline High Consump Low Tech real GDP RMB2000/Yr 6.67E+13 5.64E+13 7.23E+13 per capita real GDP RMB2000/Yr 46,829 39,745 47,580 unemployment rate % of workforce 6.12% 18.67% 0.00% total electricity demand Bn KWH/Yr 8,191 7,124 8,949 total petroleum demand MT/Yr 1,059 791 1,294 fossil fuel CO2 emission Ton/Yr 1.16E+10 8.87E+09 1.40E+10 Ha 1.15E+08 1.19E+08 1.08E+08 Ton/Yr 6.07E+11 4.61E+11 7.93E+11 Agriculture Land total water demand   Source:  “  Consumption  and  Sustainability:  A  Quantitative  Approach  Based  on  T21  China”,  presented   by  Weishuang  Qu  at  the  Transatlantic  Research  on  Policy  Modelling  Workshop  26       Available  Tools   Research   about   collaborative   software   has   been   conducted   since   the   mid   1980's,   when   computer-­‐ human   interaction,   office   automation,   and   support   for   group   work   became   the   focus   of   research   projects.  The  term  computer-­‐supported  cooperative  work  (CSCW)  was  first  used  in  1984  and  focused   on  the  support  of  small  groups  of  people.  Other  terms  are  used  as  synonyms  for  CSCW,  especially:   collaborative  computing,  computer  mediated  communication,  and  group  decision  support  systems.   CSCW  is  defined  as  a  “computer-­‐assisted  coordinated  activity  such  as  communication  and  problem                                                                                                                             26  http://www.CROSSOVER-­‐   48  |  P a g e  
  • 49.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   solving   carried   out   by   a   group   of   collaborating   individuals"   or   as   a   system,   which   “looks   at   how   groups  work  and  seeks  to  discover  how  technology  (especially  computers)  can  help  them  work".  The   term  groupware  also  stems  from  the  1980's  and  is  defined  as  “computer-­‐based  systems  that  support   groups   of   people   engaged   in   a   common   task   (or   goal)   and   that   provide   an   interface   to   a   shared   environment".  Interestingly,  some  authors  see  groupware  as  advanced  software  that  has  to  provide   awareness   support,   while   other   authors   also   understand   code   management   or   emailing   as   groupware  systems.  In  contrast  to  groupware,  CSCW  does  not  only  comprise  technological  aspects   of  collaboration,  but  also  incorporates  psychological,  social,  and  organizational  effects.     Collaborative   technologies,   especially   in   the   field   of   groupware   and   CSCW,   are   typically   classified   using   the   time-­‐space   taxonomy   which   distinguishes   between   communication   that   occurs   at   the   same   space   or   concurrently   at   different   spaces,   and   communication   that   occurs   in   the   same   time   (synchronously)   or   in   different   times   (asynchronously).   This   view   was   established   in   1988   by   R.   Johansen   (“GroupWare:   Computer   Support   for   Business   Teams”,   The   Free   Press,   New   York)   and   taken  on  in  various  related  publications.  The  following  figure  depicts  the  typical  time-­‐space  matrix  as   presented  in  these  publications.        Figure  8:  the  Time-­‐Space  Matrix     The   matrix   divides   collaborative   technologies   into   four   possible   constellations,   while   each   of   these   constellations  can  be  supported  better  or  worse  by  different  communication  media.   By  the  way  the  architecture  of  a  collaborative   modelling  tool,  i.e.,  a  system   that   supports   a   group   in   developing   models,   is   still   under   investigation.   Some   authors   have   suggested   groupware   systems   that   help   teams   in   collective   sense-­‐making   which   is   an   important   part   of   the   modelling   process.   Conklin,   Selvin,   Buckingham   and   Sierhuis   in   “Facilitated   Hypertext   for   Collective   Sensemaking:   15   Years  on  from  gIBIS”,  a  paper  presented  in  2003  during  the  8th  International  Working  Conference  on   the   Language-­‐Action   Perspective   on   Communication   Modelling   (Tilburg,   The   Netherlands),   reports   on  an  approach,  Compendium,  that  is  the  result  of  15  years  of  experience.  Compendium  combines   three   different   areas:   meeting   facilitation,   graphical   hypertext   and   conceptual   frameworks.   To   make   them   work,   facilitation   is   viewed   as   essential   to   remove   the   cognitive   overhead   for   the   group   49  |  P a g e  
  • 50.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   members.  As  groupware  systems  address  the  important  issue  of  collective  sense-­‐making  they  can  be   used   as   the   core   of   a   collaborative   modelling   tool.   So   far   these   systems   are   typically   tailored   for   specific   modelling   languages   though.   For   a   collaborative   modelling   tool   they   need   to   be   more   modular  so  that  any  modelling  language  can  be  “plugged  in”  (e.g.,  other  enterprise  or  information   systems  modelling  languages).  In  addition,  there  is  also  the  need  for  a  negotiation  component  that   facilitates  structured  arguments  and  decisions  regarding  modelling  choices.  Based  on  this  reflections   and  issues,  recently  two  tools  are  emerging:   • The   COllaborative   Modelling   Architecture,   COMA 27,   allows   group   modelling.   Any   group   member  can  work  on  the  models  whenever  it  suits  them.  Any  participant  can  contribute  in   the   way   they   can:   by   just   looking   at   proposals   and   commenting   them,   by   making   minor   changes  to  them  or  maybe  even  by  making  their  own  proposals.  The  facilitator  can  see  the   status   of   the   modelling   process   at   any   time   and   can   decide   whether   a   certain   proposal   should   be   adopted   or   needs   improvement   based   on   the   comments   by   the   other   group   members  and  his  own  judgment.                                      Figure  9:  COMA,  COllaborative  Modelling  Architecture   COMA's   design   has   been   inspired   by   theoretical   insights   from   organizational   semiotics   and   driven  by  observations  of  group  modelling  behavior.  The  tool  is  implemented  in  Visual  C++   2005  on  Windows  based  on  the  UML  Pad  and  with  the  wxWidgets  GUI  library28.       • The   OCOPOMO   eParticipation   platforms,   deployed   by   Open   Collaboration   for   Policy   Modelling  FP7  project,  that  will  end  in  December  2012.                                                                                                                               27   28   50  |  P a g e  
  • 51.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP                                              Figure  10:  OCOPOMO  eParticipation  Platform   The  platform  is  a  suite  of  ICT  tools  for:   o Iterative  development  of  policies  in  a  form  of  narrative  scenarios;   o Policy  modelling,  creation  of  agent-­‐based  formal  policy  models;   o Open  and  transparent  collaboration  in  the  process  of  policy  development;   o Seamless,  goal-­‐oriented  information  exchange  between  all  the  stakeholders  (policy   analysts,  operators,  decision  makers,  wider  interest  groups,  general  public,  etc.);   o Simulation  and  visualisation  of  policy  alternatives  and  their  consequences;   A   First   prototype   was   released   in   autumn   2011   and   tested   on   a   1st   round   of   pilot   applications   started   on   winter   2011.   2nd   pilot   applications   and   evaluation  started   in   autumn   2012,  and  the  platform  has  been  released  in  December  2012.     Key  challenges  and  gaps   This  research  challenge  is  connected  to  the  research  on  Web  2.0  and  the  next  generation  web.  As  far   as  the  Policy  Modelling  in  Governance  is  concerned,  this  research  challenge  bridges  the  gap  between   citizens   and   decision   makers.   It   permits   an   early   stage   evaluation   of   the   decision   maker   mental   models   by   opening   a   dialogue   with   citizens   and   allows   for   an   exchange   of   perspectives.   It   finally   enables   the   collaboration   in   the   public   policy   modelling   process   with   the   use   of   a   rigorous   and   formal  scientific  process.     Current  research   According  to  current  research,  the  following  issues  are  being  explored:   • Group   model   building   and   systems   thinking,   focusing   on   models   when   tackling   a   mix   of   interrelated   strategic   problems   to   enhance   team   learning,   foster   consensus,   and   create   commitment;   although   people   have   different   views   of   the   situation   and   define   problems   differently,  this  current  field  of  research  shows  that  this  can  be  very  productive  if  and  when   people  learn  from  each  other  in  order  to  build  a  shared  perspective.     51  |  P a g e  
  • 52.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Web   2.0   tools   for   collaboration,   as   recently   pointed   out   in   the   FP7   project   OCOPOMO   (Open   Collaboration   in   Policy   Modelling),   which   aim   to   implement   collaborative   scenario   building   and   policy   modelling   via   an   integrated   ICT   toolbox.   OCOPOMO   provides   an   innovative   "off   the  mainstream"  bottom-­‐up  approach  to  policy  development,  combined  with  advanced  ICT   tools  and  techniques  supporting  open  collaboration.  The  project  is  developing  an  ICT-­‐based   environment   integrating   lessons   and   practical   techniques   from   complexity   science,   agent   based   social   simulation,   foresight  scenario  analysis   and  stakeholder   participation   in  order  to   formulate   and   monitor   social   policies   to   be   adopted   at   several   levels.   The   project   is   co-­‐ funded   by   the   European   Commission   under   the   7th   Framework   Program,   Theme   7.3   (ICT   for   Governance  and  Policy  Modelling).     Future  research   Future  research  should  therefore  focus  on:   • Collaborative   Internet-­‐based   modelling   tools,   allowing   more   than   one   modeller   to   cooperate,  at  the  same  time,  on  a  single  model.   • Definition   of   frameworks   allowing   even   “low-­‐skilled”   citizens   to   provide   their   contribution   (even  if  in  a  discursive  way)  to  the  modelling  process.   • Design  of  more  intuitive  and  accessible  Human-­‐Computer  Interfaces.   52  |  P a g e  
  • 53.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     3.1.3. Easy  Access  to  Information  and  Knowledge  Creation     Introduction  and  definition   According  to  a  cybernetic  view  of  intelligent  organisations  knowledge  supersedes  1.  the  facts,  2.  data   (statements  about  facts)  and  3.  meaningful  information  (what  changes  us),  the  last  also  defined  as   “the  difference  that  makes  the  difference”.  Knowledge  most  often  defined  as  “whatever  is  known,   the   body   of   truth,   information   and   principles   acquired”   by   a   subject   on   a   certain   topic.   Therefore   knowledge   is   always   embodied   in   someone.   It   implies   insight,   which,   in   turn,   enables   orientation,   and   thus   may   be   also   use   as   a   potential   for   action   (when   we   are   able   to   use   information   in   a   certain   environment,   then   we   start   to   learn,   which   is   the   process   that   helps   developing   and   grounding   knowledge).  Two  more  concepts  come  after  knowledge  on  the  same  scale,  and  are  Understanding   and   Wisdom.   Understanding   is   the   ability   to   transform   knowledge   into   effective   action,   i.e.   in-­‐depth   knowledge,  involving  both  deep  insights  into  patterns  of  relationships  that  generate  the  behaviour   of  a  system  and  the  possibility  to  convey  knowledge  to  others,  whereby  wisdom  is  a  higher  quality  of   knowledge  and  understanding  the  ethical  and  aesthetic  dimensions.   The  research  challenge  is  related  to  the  elicitation  of  information  which,  in  turn,  during  the  overall   model  building  and  use  processes  will  help  decision  makers  to  learn  how  a  certain  system  works  and   ultimately   to   gain   insights   (knowledge)   and   understanding   (apply   the   extracted   knowledge   from   those   processes)   in   order   to   successfully   implement   a   desired   policy.   It   is   important   to   note   that   other  research  fields  (in  particular,  ICT  disciplines)  tend  to  misuse  the  word  “knowledge”  and  invert   it  with  ”information”.       Why  it  matters  in  governance   Proper  information  acquisition  and  knowledge  development  are  the  key  aspect  in  all  research  fields,   so   this   research   challenge   has   a   horizontal   importance   for   research   in   general.   According   to   the   general   need   for   policy   assessment   and   evaluation,   there   are   some   specific   issues   stemming   from   this  research  challenge,  which  are  strongly  related  to  governance:   • Public  data  use  and  thus  public  information  elicitation  (by  citizens)   •  Citizens’  behavioural  data  which  are  gradually  becoming  essential  for  any  policy  assessment   process   • Interoperability  of  public  IT  systems   • Creation  of  a  common  understanding  on  a  certain  system’s  behaviour  (by  means  of  learning)   in   order   to   develop   a   shared   vision   on   the   problems   that   a   certain   policy   might   want   to   overcome     Current  Practice  and  Inspiring  cases   In   current   practice,   information   is   drawn   from   data   stored   in   different   types   of   media   (mainly   DBMS/ERPs).  Web  2.0  has  further  transformed  the  way  we  create  data  and  elicit  information  from   data.  Data  availability  ceased  to  pose  problems  as  a  result  of:   • The  Internet  growth  and  its  uptake   • User  Generated  Content  in  Social  Networks   53  |  P a g e  
  • 54.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Cooperation   of   IT   systems   from   different   organisations   thanks   to   the   Service-­‐Oriented   Architectures   (even   among   old   legacy   systems),   which   resulted   also   in   private   data   availability   • Public  Administration  Transparency  and  Public  Data  use/reuse     Available  Tools   A  review  of  the  available  tools  is  to  be  finalized.     Key  challenges  and  gaps   The  knowledge  is  still  mostly  created  and  passed  on  by  formal  methods  of  teaching,  even  though  the   advents   of   the   e-­‐Learning,   m-­‐Learning   and   webinar   fields   allow   for   an   increased   possibility   to   perform   Distance   Learning   on   the   Web.   But,   since   knowledge   is   developed   and   grounded   by   the   learning   process   through   action   in   the   environment,   the   learning   in   real   life   comes   from   committing   mistakes.  In  the  field  of  real  life  governance,  it  entails  implementing  a  wrong  policy  and  observing   the   positive   and   negative   consequences   that   this   policy   generates   (for   example   due   to   a   system’s   “policy  resistance”).  Learning  of  successes  is  also  important,  as  the  A.I.  method  is  based  on  positive   psychology.   At   present,   thanks   to   the   increasing   data   availability,   information   elicitation   process   is   much   easier,   either   by   tacitly   bringing   users   (data   generators)   to   provide   data   in   a   guided   way   (according   to   a   pre-­‐set   framework   for   data   input)   or   with   a   help   of   a   specific   process   (e.g.:   consultations  in  e-­‐Participation  tools).     Current  research   According   to   current   research,   the   main   focus   is   put   on   the   Knowledge   Management   field   or   also   (more  properly,  as  in  our  case)  to  the  Knowledge  Elicitation  field.  The  latter  basically  encompasses   the  following  steps:   • Data  retrieval  and  extraction   • Data  analysis  and  interpretation  (which  usually  produces  information)   • Data/information   adaptation   and   integration   (this   is   particularly   the   case   where   information   needs  to  be  used  in  a  model)     Future  research   There  is  still  a  large  field  to  be  explored  –  the  methods  of  extraction  of  meaningful  information  from   unstructured   sources   of   data,   e.g.   when   analysing   free   texts,   which   applies   to   all   sources   of   User-­‐ Generated  Content  (forums,  wikis,  social  networks,  etc.),  where  the  semantic  dimension  is  essential   to   derive   meaningful   information   rather   than   just   quantitatively   analysing   the   syntax   of   text.   In   general,  a  lot  of  data  is  generated  by  citizens  and  particularly  by  their  behaviour  online,  so  that  the   available   aggregated   data   sets   contains   information   on   what   a   citizen   does,   what   s/he   likes,   how   s/he   behaves   in   certain   environments,   and   so   on.   This   data   is   considered   very   valuable   both   for   private  and  public  organisations  (even  though  under  privacy  restrictions  which  have  to  be  properly   addressed).     Also,  according  to  the  knowledge  creation  and  development  of  understanding  (regarding  a  specific   system),  there  is  some  research  currently  carried  out  on  how  to  improve  the  learning  process  via  the   use   of   e-­‐Learning   systems.   In   this   respect,   it   is   crucial   to   boost   the   research   on   micro-­‐worlds,   i.e.   54  |  P a g e  
  • 55.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   complex  virtual  environments  where  reality  is  somehow  reproduced  and  where  a  decision  maker  is   trained   in   order   to   implement   his/her   strategies   and   hypothesis   and   perform   what-­‐if   analysis   without  the  need  to  necessarily  learn  from  mistakes  in  real  life.     Future  research  will  thus  have  to  focus  on  the  following  issues:   • Information   elicitation   by   analysing   and   interpreting   data,   also   taking   into   account   the   semantic  point  of  view.   • Creation   of   proper   micro-­‐worlds   (or   ILEs,   Interactive   Learning   Environments),   where   the   acquired   information   on   a   certain   system   is   used   (by   means   of   actions),   and   knowledge   is   developed   by   observation   of   the   outcomes   of   the   actions.   Also,   ILEs   will   have   to   be   integrated   into   LMS   (Learning   Management   Systems)   in   order   to   extend   the   potential   of   distance  learning  practices,  eventually  also  in  a  cooperative  way  (mass  learning).   • Interoperability  of  data  sources  in  order  to  integrate/aggregate  different  types  of  data  and   be  able  to  automatically  infer  information  from  more  meaningful  datasets.   • In  view  of  the  “Internet  of  Things”,  the  provision  of  “portable”  models/tools  for  citizens  in   order   to   gather   valuable   data   based   on   citizens’’   real   behaviours.   Moreover,   these   models   and  tools  would  enable  citizens  to  check  the  results  of  their  actions  by  analysing  in  real-­‐time   the   response   of   the   model   to   the   information   they   are   contributing   to   generate,   and   thus   evaluating   the   eventual   benefits   they   are   receiving   from   their   virtuous   behaviour   or   harm   they  are  creating  either  to  their  environment  or  to  themselves  (e-­‐Cognocracy).             55  |  P a g e  
  • 56.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   3.1.4. Model  Validation     Introduction  and  definition   Policy   makers   need   and   use   information   stemming   from   simulations   in   order   to   develop   more   effective   policies.   As   citizens,   public   administration   and   other   stakeholders   are   affected   by   decisions   based  on  these  models,  the  reliability  of  applied  models  is  crucial.  Model  validation  can  be  defined   as   ”substantiation   that   a   computerised   model   within   its   domain   of   applicability   possesses   a   satisfactory  range  of  accuracy  consistent  with  the  intended  application  of  the  model”  (Schlesinger,   1979).   Therefore,   a   policy   model   should   be   developed   for   a   specific   purpose   (or   context)   and   its   validity  determined  with  respect  to  that  purpose  (or  context)29.  If  the  purpose  of  such  a  model  is  to   answer  a  variety  of  questions,  the  validity  of  the  model  needs  to  be  determined  with  respect  to  each   question.  A  model  is  considered  valid  for  a  set  of  experimental  conditions  if  the  model’s  accuracy  is   within   its   acceptable   range,   which   is   the   amount   of   accuracy   required   for   the   model’s   intended   purpose.   The   substantiation   that   a   model   is   valid   is   generally   considered   to   be   a   process   and   is   usually   part   of   the   (total)   policy   model   development   process     (Sargent,   2008).   For   this   purpose,   specific  and  integrated  techniques  and  ICT  tools  are  required  to  be  developed  for  policy  modelling.     Model  validation  is  composed  of  two  main  phases:   • Conceptual  model  validation,  i.e.  determining  that  theories  and  assumptions  underlying  the   conceptual  model  are  correct  and  that  the  model’s  representation  of  the  problem  entity  and   the   model’s   structure,   logic,   and   mathematical   and   causal   relationships   are   “reasonable”   for   the  intended  purpose  of  the  model.   • Computerised  model  verification  ensures  that  computer  programming  and  implementation   of   the   conceptual   model   are   correct,   as   well   as   states   that   the   overall   behaviour   of   the   model  is  in  line  with  the  available  historical  data.     Why  it  matters  in  governance   Model  Validation  is  connected  both  to  modelling  and  simulation.  According  to  the  general  need  for   policy   assessment   and   evaluation,   there   are   some   specific   issues   stemming   from   the   Model   Validation,  which  are  strongly  related  to  governance:   • Reliability  of  models:  policy  makers  use  simulation  results  to  develop  effective  policies  that   have  an  important  impact  on  citizens,  public  administration  and  other  stakeholders.  Model   validation  is  fundamental  to  guarantee  that  the  output  (simulation  results)  for  policy  makers   is  reliable.   • Acceleration   of   policy   modelling   process:   policy   models   must   be   developed   in   a   timely   manner   and   at   minimum   cost   in   order   to   efficiently   and   effectively   support   policy   makers.   Model   validation   is   both   cost   and   time   consuming   and   should   be   automated   and   accelerated.                                                                                                                             29  Some  researchers  claim  that  the  category  "validity"  has  little  meaning  in  relation  to  policy  models,  as  they  are  generally  a   form   of   narrative   or   storytelling   so   that   their   value   comes   in   the   act   of   obtaining   a   better   understanding   of   the   system   and   being  able  to  communicate  concepts  effectively  and  spur  discussion  between  different  stakeholders.   56  |  P a g e  
  • 57.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Modular  and  re-­‐usable  models:  a  policy  model  developer  deciding  to  re-­‐use  existing  models   or   compose   them,   stumble   across   the   issue   of   models’   reliability.   Model   validation   can   be   used  for  certifying  this  reliability  and  creating  a  database  of  validated  models.     Current  Practice  and  Inspiring  cases   In   current   practice   the   most   frequently   used   is   a   decision   of   the   development   team   based   on   the   results   of   the   various   tests   and   evaluations   conducted   as   part   of   the   model   development   process.   Another   approach   is   to   engage   users   in   the   validation   process.   When   developing   large-­‐scale   simulation  models,  the  validation  of  a  model  can  be  carried  by  an  independent  third-­‐party.  Needless   to  say,  that  the  third  party  needs  to  have  a  thorough  understanding  of  the  intended  purpose  of  the   simulation   model.   Finally,   the   scoring   model   can   be   used   for   testing   the   model’s   validity   (e.g.   see   Balci   1989;   Gass   1983;     Gass   &   Joel   1987).   Scores   (or   weights)   are   determined   subjectively   when   conducting   various   aspects   of   the   validation   process   and   then   combined   to   determine   category   scores   and   an   overall   score   for   the   simulation   model.   A   simulation   model   is   considered   valid   if   its   overall  and  category  scores  are  greater  than  some  passing  score.     Available  Tools   A  review  of  the  available  tools  is  to  be  finalized.     Key  challenges  and  gaps   Typically   all   above-­‐mentioned   approaches   are   applied   after   the   simulation   model   has   been   developed.   Performing   a   complete   validation   effort   after   the   simulation   model   has   been   finalised   requires   both   time   and   money.   However,   conducting   model   validation   concurrently   with   the   development  of  the  simulation  model  enables  the  model  development  team  to  receive  inputs  earlier   on   each   stage   of   model   development.   Therefore,   ICT   tools   for   speeding   up,   automating   and   integrating   model   validation   process   into   policy   model   development   process   are   necessary   to   guarantee  the  validity  of  models  with  an  effective  use  of  resources.     Current  research   In  Current  research,  there  are  a  large  number  of  subjective  and  objective  validation  techniques  used   for   verifying   and   validating   the   modules   and   the   overall   model.   Robert   G.   Sargent   at   the   Syracuse   University     in   2010   provided   a   relevant   ones:   Animation;   Comparison   to   Other   Model;   Degenerate   Tests;   Event   Validity;   Extreme   Condition   Tests;   Face   Validity;   Historical   Data   Validation;   Historical   Methods;   Internal   Validity;   Multistage   Validation;   Operational   Graphics;   Parameter   Variability   /   Sensitivity  Analysis;  Predictive  Validation;  Traces;  and  Turing  Tests.  Furthermore,  he  described  a  new   statistical  procedure  for  validating  simulation  and  analytic  stochastic  models  using  hypothesis  testing   when   the   amount   of   model   accuracy   is   specified.   This   procedure   provides   for   the   model   to   be   accepted   if   the   difference   between   the   system   and   the   model   outputs   are   within   the   specified   ranges  of  accuracy.  The  system  must  be  observable  to  allow  data  to  be  collected  for  validation.       Future  research   Future  research  should  explore  the  following  issues:   • In   order   to   speed   up   and   reduce   the   cost   of   a   model   validation   process,   user-­‐friendly   and   collaborative   statistical   software   should   be   developed,   possibly   combined   with   expert   systems  and  artificial  intelligence.   57  |  P a g e  
  • 58.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Due   to   the   big   gap   between   theory   and   practice,   the   considerable   opportunity   exists   for   the   study   and   application   of   rigorous   verification   and   validation   techniques.   In   the   current   practice,  the  comparison  of  the  model  and  system  performance  measures  is  typically  carried   out  in  an  informal  manner.   • Complicated  simulation  models  are  usually  either  not  validated  at  all  or  are  only  subjectively   validated;   for   example,   animated   output   is   eyeballed   for   a   short   while.   Therefore,   complexity  issues  in  model  validation  may  be  better  addressed  through  the  development  of   more  suitable  methodologies  and  tools.   • Model   validation   is   not   a   discrete   step   in   the   simulation   process.   It   needs   to   be   applied   continuously   from   the   formulation   of   the   problem   to   the   implementation   of   the   study   findings   as   a   completely   validated   and   verified   model   does   not   exist.   Validation   and   verification  process  of  a  model  is  never  completed.   • As  the  model  developers  are  inevitably  biased  and  may  be  concentrated  on  positive  features   of  the  given  model,  the  third  party  approach  (board  of  experts)  seems  to  be  a  better  solution   in  model  validation.   • Considering  the  ranges  that  simulation  studies  cover  (from  small  models  to  very  large-­‐scale   simulation   models),   further   research   is   needed   to   determine   with   respect   to   the   size   and   type  of  simulation  study     o o How  should  model  validation  be  managed,     o • Which  model  validation  approach  should  be  used,     What  type  of  support  system  software  for  model  validation  is  needed.   Validating   large-­‐scale   simulations   that   combine   different   simulation   (sub-­‐)   models   and   use   different  types  of  computer  hardware  such  as  in  currently  being  done  in  HLA  (Higher  Level   Architecture).  A  number  of  these  VV&A  issues  need  research,  e.g.  how  does  one  verify  that   the  simulation  clocks  and  event  (message)  times  (timestamps)  have  the  same  representation   (floating   point,   word   size,   etc.)   and   validate   that   events   having   time   ties   are   handled   properly.                             58  |  P a g e  
  • 59.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP                                                                                 3.1.5. Immersive  Simulation     Introduction  and  definition   As  policy  models  grow  in  size  and  complexity,  the  process  of  analysing  and  visualising  the  resulting   large   amounts   of   data   becomes   an   increasingly   difficult   task.   Traditionally,   data   analysis   and   visualisation   were   performed   as   post-­‐processing   steps   after   a   simulation   had   been   completed.   As   simulations   increased   in   size,   this   task   became   increasingly   difficult,   often   requiring   significant   computation,   high-­‐performance   machines,   high   capacity   storage,   and   high   bandwidth   networks.   Computational   steering   is   an   emerging   technology   that   addresses   this   problem   by   “closing   the   loop”   and   providing   a   mechanism   for   integrating   modelling,   simulation,   data   analysis   and   visualisation.   This   integration   allows   a   researcher   to   interactively   control   simulations   and   perform   data   analysis   while   avoiding   many   of   the   pitfalls   associated   with   the   traditional   batch   /   post   processing   cycle.   This   research   challenge   refers   to   the   issue   of   the   integration   of   visualisation   techniques   within   an   integrated   simulation   environment.   This   integration   plays   a   crucial   role   in   making   the   policy   modelling   process   more   extensive   and,   at   the   same   time,   comprehensible.   In   fact,   the   real   aim   of   interactive   simulation   is,   on   the   one   hand,   to   allow   model   developers   to   easily   manage   complex   models   and   their   integration   with   data   (e.g.   real-­‐time   data   or   qualitative   data   integration)   and,   on   the  other  hand,  to  allow  the  other  stakeholders  not  only  to  better  understand  the  simulation  results,   but   also   to   understand   the   model   and,   eventually,   to   be   involved   in   the   modelling   process.   Interactive   simulation   can   dramatically   increase   the   efficiency   and   effectiveness   of   the   modelling   and   simulation   process,   allowing   the   inclusion   and   automation   of   some   phases   (e.g.   output   and   feedback  analysis)  that  were  not  managed  in  a  structured  way  up  to  this  point.     Why  it  matters  in  governance   Immersive   simulation   is   a   particular   aspect   of   simulation.   As   far   as   the   Policy   Assessment   in   Governance  is  concerned,  this  challenge  may:   • Accelerate  the  simulation  process:  policy  makers  would  be  able  to  analyse  simulation  results,   eventually  run  new  scenarios  and  make  decisions  as  soon  as  possible  and  at  the  minimum   cost.   •  Collaborative   environment:   the   bigger   is   the   number   of   stakeholders   involved   in   policy   modelling   and   simulation   process,   the   greater   is   the   necessity   of   an   interactive   simulation   environment   that   allows   non-­‐experts   to   use   the   model   and   understand   results   as   well   as   permit  experts  to  easily  understand  new  requirements  and  consequent  modification.   • Citizen   engagement:   interactive   simulation   tools   help   to   engage   citizens   in   policy-­‐making   process  and  to  display  to  them  in  a  simple  way  the  results.   • Data  integration:  interactive  simulation  tools  allow  better  managing  of  a  large  number  and   different  types  of  data  and  information,  both  for  input  and  output/feedback  analysis.       Current  Practice  and  Inspiring  cases   In   current   practice,   data   analysis   and   visualisation,   albeit   critical   for   the   process,   are   often   performed  as  a  post-­‐processing  step  after  batch  jobs  are  run.  For  this  reason,  the  errors  in  validating   59  |  P a g e  
  • 60.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   the   results   of   the   entire   simulation   may   be   discovered   only   during   post-­‐processing.   What   is   more,   the  decoupling  of  simulation  and  analysis/visualisation  can  present  serious  scientific  obstacles  to  the   researcher   in   interpreting   the   answers   to   “what   if”   questions.   Given   the   limitations   of   the   batch   /   post   processing   cycle,   it   might   be   advisable   to   break   the   cycle   and   improve   the   integration   of   simulation   and   visualisation.   Implementation   of   an   interactive   simulation   and   visualisation   environment   requires   a   successful   integration   of   the   many   aspects   of   scientific   computing,   including   performance   analysis,   geometric   modelling,   numerical   analysis,   and   scientific   visualisation.   These   requirements   need   to   be   effectively   coordinated   within   an   efficient   computing   environment.   Recently,   several   tools   and   environments   for   computational   steering   have   been   developed.   They   range   from   tools   that   modify   performance   characteristics   of   running   applications,   either   by   automated   means   or   by   user   interaction,   to   tools   that   modify   the   underlying   computational   application,  thereby  allowing  application  steering  of  the  computational  process.       Available  Tools   A  review  of  the  available  tools  is  to  be  finalized.     Key  challenges  and  gaps   The  development  of  immersive  tools  is  still  based  on  model  developers  needs  and  therefore  a  gap   still   exists   between   requirements   of   policy   makers   and   those   of   developers.   In   a   collaborative   modelling  environment,  interaction  is  fundamental  in  order  to  speed  up  the  process  and  make  ICT   tools  user-­‐friendly  for  all  the  stakeholders  involved  in  the  policy  model  development  process.       Current  research   In  the  current  research,  interactive  visualisation  typically  combines  two  main  approaches:  providing   efficient  algorithms  for  the  presentation  of  data  and  providing  efficient  access  to  the  data.  The  first   advance  is  evident  albeit  challenging.  Even  though  computers  continually  get  faster,  data  sizes  are   growing   at   an   even   more   rapid   rate.   Therefore,   the   total   time   from   data   to   picture   is   not   decreasing   for  many  of  the  problem  domains.  Alternative  algorithms,  such  as  ray  tracing    (Nakayama,  2002)  and   view   dependent   algorithms     (Lessig,   2009)   can   restore   a   degree   of   interactivity   for   very   large   datasets.   Each   of   those   algorithms   has   its   trade-­‐offs   and   is   suitable   for   a   different   scenario.   The   second   advance   is   less   evident   but   very   powerful.   Through   the   integration   of   visualisation   tools   with   simulation   codes,   a   scientist   can   achieve   a   new   degree   of   interactivity   through   the   direct   visualisation   and   even   manipulation   of   the   data.   The   scientist   does   not   necessarily   wait   for   the   computation  to  finish  before  interacting  with  the  data,  but  can  interact  with  a  running  simulation.   While  conceptually  simple,  this  approach  poses  numerous  technical  challenges.     Future  research   With  regard  to  future  research,  interactive  simulation  plays  a  crucial  role  in  a  collaborative  modelling   environment.  The  trade-­‐off  between  the  possibility  of  enlarging  models  and  including  several  kinds   of   data,   and   the   number   of   people   that   can   understand   and   modify   the   model   should   be   deeply   analysed.  For  this  purpose,  some  fundamental  issues  must  be  approached:   • Systems  should  be  modular  and  easy  to  extend  within  the  existing  codes.   • Users  of  the  systems  should  be  able  to  add  new  capabilities  easily  without  being  experts  in   systems  programming.   60  |  P a g e  
  • 61.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Input  /  output  systems  should  be  easily  integrated.   • Steering   systems   should   be   adaptable   to   hardware   ranging   from   the   largest   of   supercomputing  systems  to  low-­‐end  workstations  and  PCs.   3.1.6. Output  Analysis  and  Knowledge  Synthesis     Introduction  and  definition   Inputs   driving   a   simulation   are   often   random   variables,   and   because   of   this   randomness   in   the   components   driving   simulations,   the   output   from   a   simulation   is   also   random,   so   statistical   techniques   must   be   used   to   analyse   the   results.   In   particular,   the   output   processes   are   often   non-­‐ stationary  and  auto-­‐correlated  and  classical  statistical  techniques  based  on  independent  identically   distributed  observations  are  not  directly  applicable.  In  addition,  by  observing  a  simulation  output,  it   is  possible  to  infer  the  general  structure  of  a  system,  so  ultimately  gaining  insights  on  that  system   and   being   able   to   synthesise   knowledge   on   it.   There   is   also   the   possibility   to   review   the   initial   assumptions  by  observing  the  outcome  and  by  comparing  it  to  the  expected  response  of  a  system,   i.e.   performing   a   modelling   feedback   on   the   initial   model.   Finally,   one   of   the   most   important   uses   of   simulation   output   analysis   is   the   comparison   of   competing   systems   or   alternative   system   configurations.   Visualisation  tools  are  essentials  for  the  correct  execution  of  this  iterative  step.  The  present  research   challenge   deals   with   the   issue   of   output   analysis   of   a   policy   model   and,   at   the   same   time,   of   feedback  analysis  in  order  to  incrementally  increase  and  synthesise  the  knowledge  of  the  system.     Why  it  matters  in  governance   Output   analysis   is   a   specific   aspect   of   simulation.   According   to   the   general   need   for   policy   assessment  and  evaluation,  there  are  some  specific  issues  stemming  from  the  output  analysis,  which   are  strongly  related  to  governance:   • Acceleration   of   policy   assessment   process:   automated   output   analysis   tools   would   help   policy  makers  to  efficiently  and  effectively  analyse  the  impacts  of  a  policy  even  if  the  large   number  of  simulation  data  must  be  taken  into  account   • Citizen   engagement:   user-­‐friendly   automated   tools   for   output   analysis   can   be   offered   to   citizens   in   order   to   share   the   simulation   results   and   better   engage   them   in   policy-­‐making   process.     Current  Practice  and  Inspiring  cases   In   the   current   practice   a   large   amount   of   time   and   financial   resources   are   spent   on   model   development   and   programming,   but   little   effort   is   allocated   to   analyse   the   simulation   output   data   in   an   appropriate   manner.   As   a   matter   of   fact,   a   very   common   way   of   operating   is   to   make   a   single   simulation   of   somewhat   arbitrary   length   run   and   then   treat   the   resulting   simulation   estimates   as   being   the   "true"   characteristics   of   the   model.   Since   random   samples   from   probability   distributions   are   typically   used   to   drive   a   simulation   model   through   time,   these   estimates   are   realisations   of   random   variables   that   may   have   large   variances.   As   a   result,   these   estimates   could,   in   a   particular   simulation  run,  differ  greatly  from  the  corresponding  true  answers  for  the  model.  The  net  effect  is   that  there  may  be  a  significant  probability  of  making  erroneous  inferences  about  the  system  under   study.   Historically,   there   are   several   reasons   why   output   data   analysis   was   not   conducted   in   an   appropriate   manner.   First,   users   often   have   the   unfortunate   impression   that   simulation   is   just   an   61  |  P a g e  
  • 62.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   exercise   in   computer   programming.   Consequently,   many   simulation   studies   begun   with   heuristic   model   building   and   computer   coding,   and   end   with   a   single   run   of   the   program   to   produce   "the   answers."  In  fact,  however,  a  simulation  is  a  computer-­‐based  statistical  sampling  experiment.  Thus,   if  the  results  of  a  simulation  study  are  to  have  any  meaning,  appropriate  statistical  techniques  must   be   used   to   design   and   analyse   the   simulation   experiments   and   ICT   tools   must   be   developed   to  make   the   process   more   effective   and   efficient.   In   addition,   there   are   some   important   issues   of   output   analysis  that  are  not  strictly  connected  to  statistics.  In  particular,  an  evident  gap  in  literature  regards   the  analysis  and  integration  of  feedbacks  in  modelling  and  simulation  process.  Actually,  stakeholders   are   involved,   in   a   post-­‐processing   phase,   in   order   to   analysis   the   results   (more   often   only   the   elaboration  of  them)  and  understand  something  about  the  policy.  Sometimes  they  are  able  to  give  a   feedback   on   the   difference   between   their   expectations   and   the   result   but   the   process   is   not   structured   and   effective   tools   are   lacking.   The   development   of   tools   for   analysing   and   integrating   feedbacks  should  be  explored  in  order  to  enlarge  the  number  of  stakeholders  involved  and,  at  the   same  time,  to  allow  efficient  and  effective  modification  at  each  phase  of  the  process,  incrementally   increasing  the  knowledge  of  the  model  and,  consequently,  of  the  given  policy.     Available  Tools   A  review  of  the  available  tools  is  to  be  finalized.     Key  challenges  and  gaps   A  fundamental  issue  for  statistical  analysis  is  that  the  output  processes  of  virtually  all  simulations  are   non-­‐stationary   (the   distributions   of   the   successive   observations   change   over   time)   and   auto   correlated   (the   observations   in   the   process   are   correlated   with   each   other).   Thus,   classical   statistical   techniques  based  on  independent  identically  distributed  observations  are  not  directly  applicable.  At   present,   there   are   still   several   output-­‐analysis   problems   for   which   there   is   no   commonly   accepted   solution,   and   the   solutions   that   are   available   are   often   too   complicated   to   apply.   Another   impediment   to   obtaining   accurate   estimates   of   a   model's   true   parameters   or   characteristics   is   the   cost   of   the   computer   time   needed   to   collect   the   necessary   amount   of   simulation   output   data.   Indeed,  there  are  situations  where  an  appropriate  statistical  procedure  is  available,  but  the  cost  of   collecting  the  amount  of  data  dictated  by  the  procedure  is  prohibitive.           Current  research   In  current  research,  main  references  are  Law  (1983),  Nakayama  (2002),    Alexopoulos  &  Kim  (2002),     Goldsman   &   Tokol   (2000),   Kelton   (1997),   Alexopoulos   &   Seila   (1998),   Goldsman   &   Nelson   (1998),     Law  (2006).   For  output  analysis,  there  are  two  types  of  simulations:   • Finite-­‐horizon  simulations.  In  this  case,  the  simulation  starts  in  a  specific  moment  and  runs   until   a   terminating   event   occurs.   The   output   process   is   not   expected   to   achieve   steady-­‐state   behaviour   and   any   parameter   estimated   from   the   output   will   be   transient   in   a   sense   that   its   value   will   depend   upon   the   initial   conditions   (e.g.   a   simulation   of   a   vehicle   storage   and   distribution  facility  in  a  week  time).   • Steady-­‐state  simulations.  The  purpose  of  a  steady-­‐state  simulation  is  the  study  of  the  long-­‐ run   behaviour   of   the   system   of   interest.   A   performance   measure   of   a   system   is   called   a   62  |  P a g e  
  • 63.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   steady-­‐state   parameter   if   it   is   a   characteristic   of   the   equilibrium   distribution   of   an   output   stochastic  process  (e.g.  simulation  of  a  continuously  operating  communication  system  where   the  objective  is  the  computation  of  the  mean  delay  of  a  data  packet).     Future  research   Referring   to   previous   cited   works   and   in   particular   to   Goldsman   (2010),   future   research   should   further  explore  following  issues:   • • Allowing  an  incremental  understanding  of  the  model  (knowledge  synthesis)   • Adapting  Design  Of  Experiment  (DOE)  for  policy  model  simulation   •                                                                       ICT  tools  for  supporting  or  automating  output/feedback  analysis   Use  and  integration  of  more-­‐sophisticated  variance  estimators   • Better  ranking  and  selection  techniques.   63  |  P a g e  
  • 64.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP         3.2. Data-­‐powered  Collaborative  Governance   3.2.1. Big  Data     Summary  Overview     Current  free  tools   The   freely   available   tools   permit   to   overcome   data   limitations,   simplify   the   analytical   process   and   visualize   results.  The  functionalities  provided   by  these  software  are:   -­‐Massively  parallel  processing  (MPP)   database   product   for   large-­‐scale   analytics   and   next-­‐gen   data   warehousing   -­‐Data-­‐parallel   implementations   of   statistical   and   machine   learning   methods   -­‐Visual  data  mining  modelling       Top  market  tools   Current  and  Future  Research   -­‐Data  storage  platforms  and  other   -­‐Technologies   for   collecting   cleaning,   information   infrastructure   storing   and   managing   data:   data   solutions   warehouse;   pivotal   transformation;   ETL;   I/O;   efficient   archiving,   storing,   indexing,   -­‐Massive   parallel   processing   retrieving,   and   recovery;   streaming,   (MPP)   filtering,   compressed   sensing   sufficient   -­‐Dataflow   engines,   software   statistics;  automatic  data  annotation;  Large   Database   Management   Systems;   storage   interconnect  technologies   architectures;   data   validity,   integrity,   -­‐Data   discovery   and   exploration   consistency,   uncertainty   management;   tools   languages,   tools,   methodologies   and   -­‐Built-­‐in  text  analytics,  enterprise-­‐ programming  environments   grade   security   and   administrative   -­‐Technologies   for   summarizing   data   and   tools   extracting   some   meaning:   reports;   -­‐Real-­‐time   analytics   processing   dashboard;   statistical   analysis   and   inference;   Bayesian   techniques;   (RTAP)   information   extraction   from   unstructured,   -­‐Visualization   features   supporting   multimodal   data;   scalable   and   interactive   exploratory   and   discovery   data   visualization;   extraction   and   analytics   integration   of   knowledge   from   massive,   -­‐On-­‐line   analytical   processing   complex,   multi-­‐modal,   or   dynamic   data;   data   mining;   scalable   machine   learning;   (OLAP)   data-­‐driven   high   fidelity   simulations;   -­‐Business   intelligence   (BI),   Data   scalable   machine   learning;   predictive   Warehouse  (DW)     modelling,   hypothesis   generation   and   discovery   -­‐Enterprise   Data   Warehouse   automated   -­‐Technologies   for   using   data   a   decision   (EDW)     tool:  Decision  Trees,  Pro-­‐Con  Analysis,  Rule     Based   Systems,   Neural   Networks,   Tradeoff   based  Decisions       Introduction  and  definition   Big  Data  refers  to  dataset  that  cannot  be  stored,  captured,  managed  and  analysed  by  the  mean  of   conventional  database  software.  Thereby  Big  Data  is  a  subjective  rather  than  a  technical  definition,   because   it   does   not   involve   a   quantitative   threshold   (e.g.   in   terms   of   terabytes),   but   instead   a   moving   technological   one.   Keeping   that   in   mind,   the   definition   of   Big   Data   in   many   sectors   ranges   from   a   few   terabytes30  to   multiple   petabytes31.   The   definition   of   Big   Data   does   not   merely   involve   the   use   of   very   large   data   sets,   but   concerns   also   a   computational   turn   in   thought   and   research   (Burkholder, L, ed. 1992).                                                                                                                               30    1  terabyte  is  equal  to  1  trillion  bytes      1  petabyte  is  equal  to  1000  terabytes   31 64  |  P a g e  
  • 65.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   As   stated   by   Latour   (2009)   when   the   tool   is   changed,   also   the   entire   social   theory   going   with   it   is   different.   In   this   view   Big   Data   has   emerged   a   system   of   knowledge   that   is   already   changing   the   objects   of   knowledge   itself,   as   it   has   the   capability   to   inform   how   we   conceive   human   networks   and   community.  Big  Data  creates  a  radical  shift  in  how  we  think  research  itself.  As  argued  by  Lazer  et  al.   (2009),  not  only  we  are  offered  the  possibility  to  collect  and  analyze  data  at  an  unprecedented  depth   and  scale,  but  also  there  is  a  change  in  the  processes  of  research,  the  constitution  of  knowledge,  the   engagement   with   information   and   the   nature   and   the   categorization   of   reality.   The   potential   stemming   from   the   availability   of   a   massive   amount   of   data   is   exemplified   by   Google.   It   is   widely   believed   that   the   success   of   the   Mountain   View   company   is   due   to   its   brilliant   algorithms,   e.g.   PageRank.   In   reality   the   main   novelties   introduced   in   1998,   which   brought   to   second   generation   search   engines,   involved   the   recognition   that   hyperlinks   were   an   important   measure   of   popularity   and   the   use   of   the   text   of   hyperlinks   (anchortext)   in   the   web   index,   giving   it   a   weight   close   to   the   page   title.   This   is   because   first   generation   search   engines   used   only   the   text   of   the   web   pages,   while   Google  added  two  data  set  (hyperlinks  and  anchortext),  so  that  even  a  less  than  perfect  algorithm   exploiting   this   additional   data   would   obtain   roughly   the   same   results   as   PageRank.   Another   example   is   the   Google’s   AdWords   keyword   auction   model.   Overture   had   previously   shown   that   ranking   advertisers   for   a   given   keyword   based   purely   on   their   bids   was   an   efficient   mechanism.   Google   improved   the   tool   by   adding   the   data   on   the   clickthrough   rate   (CTR)   on   each   advertiser's   ad,   so   that   advertisers  were  ranked  by  their  bid  and  their  CTR.       Why  it  matters  in  governance   Big  Data  have  a  huge  impact  also  in  governance  and  policy  making.  In  fact  their  benefits  apply  to  a   wide  variety  of  subjects:   • Health   care:   making   care   more   preventive   and   personalized   by   relying   on   a   home-­‐based   continuous   monitoring,   thereby   reducing   hospitalization   costs   while   increasing   quality.   Detection  of  infectious  disease  outbreaks  and  epidemic  development   • Education:   by   collecting   all   the   data   on   students’   performance,   it   would   be   possible   to   design   more   effective   approaches.   The   collection   of   these   data   is   made   possible   thanks   to   massive  Web  deployment  of  educational  activities   • Urban   planning:   huge   high   fidelity   geographical   datasets   describing   people   and   places   are   generated  from  administrative  systems,  cell  phone  networks,  or  other  similar  sources.   • Intelligent  transportation  based  on  the  analysis  and  visualization  of  road  network  data,  so  as   to  implement  congestion  pricing  systems  and  reduce  traffic   • The   use   of   ubiquitous   data   collection   through   sensors   networks   in   order   to   improve   environmental  modelling     • Analysis  and  clarification  of  the  energy  pattern  use  through  data  analytics  and  smart  meters,   which  can  be  useful  for  the  adoption  of  energy  saving  policies  avoiding  blackouts   • Integrated  analysis  of  contracts  in  order  to  find  relations  and  dependencies  among  financial   institutions  in  order  to  assess  the  financial  systemic  risk   • The  analysis  of  conversation  in  social  media  and  networks,  as  well  as  the  analysis  of  financial   transaction  carried  out  by  alleged  terrorists,  which  can  be  used  for  homeland  security     • Assessment   of   computer   security   by   the   mean   of   the   logged   information   analysis,   i.e.   Security  Information  and  Event  Management   • Better  track  of  food  and  pharmaceutical  production  and  distribution  chain   65  |  P a g e  
  • 66.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Collect  data  on  water  and  sewer  usage  in  order  to  reduce  water  consumption  by  detecting   leaks   • Use  of  sensors,  GPS,  cameras  and  communication  systems  for  crisis  detection,  management   and  response     • Use  of  sensors’  data  for  carbon  footprint  management         Policy  Applications  of  Big  Data  Tools   There  is  a  growing  body  of  evidence  highlighting  the  applications  of  Big  Data  not  only  in  traditional   hard  science  and  business,  but  also  in  policy  making  due  to  the  predictive  power  of  the  data.  Let  us   see  some  applications:   • Predictability   of   human   behaviour   and   social   events.   A  research  team  from  Northwestern   University 32  was   able   to   predict   people’s   location   based   on   mobile   phone   information   generated   from   past   movements.   Moreover   Pentland   from   MIT 33  conducted   a   research   showing  that  mobile  phones  can  be  used  as  sensors  for  predicting  human  behaviour,  as  they   can  quantify  human  movements  in  order  to  explain  changes  in  commuting  patterns  given  for   example   by   unemployment.   Recently   another   research   team   from   Northeastern   University   was  able  to  predict  the  voting  outcome  in  the  scope  of  a  famous  US  television  programme   (American  Idol)  based  on  Twitter  activity  during  the  time  span  defined  by  the  TV  show  airing   and  the  voting  period  following  it34   • Public   health.   Online   data   can   be   used   for   syndromic   surveillance,   also   called   infodemiology35.   As   an   example   Google   Flu   Trends   is   a   tool   based   on   the   prevalence   of   Google  queries  for  flu-­‐like  symptoms.  As  shown  by  Ginsberg  et  al.  (2008)36  it   is   then  possible   to  use  search  queries  to  detect  influenza  epidemics  in  areas  with  a  large  population  of  web   search  users.  In  fact  according  to  the  US  Center  for  Disease  Control  and  Prevention  (CDC)37  a   great  availability  of  data  coming  from  online  queries  can  help  to  detect  epidemic  outbursts   before  laboratory  analysis.  Another  related  tool  is  the  Google  Dengue  Trend.  In  this  view  the   analysis  of  health  related  Tweets  in  US  by  Paul  and  Dredze  (2011)38  found  a  high  correlation   between   the   modeled   and   the   actual   flu   rate.   In   the   same   way   Twitter’s   data   can   be   analyzed   to   study   the   geographic   spread   of   a   virus   or   disease39.   Finally   we   can   talk   about   Healthmap 40  in   which   data   from   online   news,   eyewitness   reports,   expert-­‐curated   discussions,   official   reports,   are   used   to   get   a   thorough   view   of   the   current   global   state   of   infectious  diseases  which  is  visualised  on  a  map   • Global   food   security.   The   Food   and   Agriculture   Organization   of   the   UN   (FAO)   is   chartered   with  ensuring  that  the  world’s  knowledge  of  food  and  agriculture  is  available  to  those  who   need   it   when   they   need   it   and   in   a   form   which   they   can   access   and   use41.   In   fact   human   population   will   approach   9   billion   by   2050,   thereby   it   will   be   necessary   to   put   in   place                                                                                                                             32>   34  http://www.mobs-­‐   35­‐infodemiologyamia-­‐proc.pdf   36   apers/detecting-­‐influenza-­‐epidemics.pdf  >   37   38     39     40   41  and­‐3e69-­‐4ba0-­‐86a5-­‐b8f7c8322ebe.pdf   33 66  |  P a g e  
  • 67.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   policies   aimed   at   ensuring   a   sufficient   and   fair   distribution   of   resources.   In   fact   the   world   food  production  will  have  to  increase  by  60%  by  increasing  the  agricultural  production  and   fighting  water  scarcity.  The  online  data  portal  to  be  launched  by  FAO  will  enhance  planners’   and  decision  makers’  capacity  to  estimate  agricultural  production  potentials  and  variability   under  different  climate  and  resources  scenarios   • Environmental  analysis.  In  the  last  United  Nations  conference  on  climate  (i.e.  COP  17)  taking   place  in  2011,  The  European  Environment  Agency,  the  geospatial  software  company  Esri  and   Microsoft  presented  the  network  Eye  on  Earth42,  which  can  be  used  to  create  an  online  site   and   group   of   services   for   scientists,   researchers   and   policy   makers   in   order   to   share   and   analyze   environmental   and   geospatial   data.   Other   three   projects   launched   by   these   institutions   at   COP   17   include   WaterWatch   (using   EEA’s   water   data);   AirWatch,   (about   EEA’s   air   quality   data);   and   finally   NoiseWatch,   which   is   a   combination   between   environmental   data   with   user-­‐generated   information   provided   by   citizens.   Moreover   during   2010   United   Nations   climate   meeting   (COP   16)   Google   launched   its   own   satellite   and   mapping   service   Google   Earth   Engine43,   which   is   a   combination   of   a   computing   platform,   an   open   API   and   satellite   imagery   along   25   years.   All   these   tools   will   be   available   to   scientists,   researchers   and   governmental   agencies   for   analyzing   the   environmental   conditions   in   order   to   make   sustainability   decisions.   In   this   way   the   government   of   Mexico   created   a   map   of   the   country’s   forest   incorporating   53,000   Landsat   images,   which   can   be   used   by   the   federal   authority  and  the  NGOs  to  make  decisions  about  land  use  and  sustainable  agriculture.   • Crisis   management   and   anticipation.   In   occasion   of   the   Haiti   earthquake44:   an   European   Commission’s   Joint   Research   Center   team   used   the   damage   reports   mapped   on   the   Ushahidi-­‐Haiti   platform45  to   show   that   this   crowdsourced   data   can   help   predict   the   spatial   distribution  of  structural  damage  in  Port-­‐au-­‐Prince.  Their  model  based  on  1645  SMS  reports   crowdsourced  data  almost  perfectly  predicts  the  structural  damage  of  most  affected  areas   reported   in   the   World   Bank-­‐UNOSAT-­‐JRC   damage   assessment   performed   by   600   experts   from  23  countries  in  66  days  based  on  high  resolution  aerial  imagery  of  structural  damage.   As  for  future  developments,  some  researches46  highlight  the  fact  that  Big  Data  can  be  used   for   crisis   management   and   anticipation   by   building   up   crisis   observatories,   i.e.   laboratories   devoted   to   the   collecting   and   processing   of   enormous   volumes   of   data   on   both   natural   systems   and   human   techno-­‐socio-­‐economic   systems,   so   as   to   gain   early   warnings   of   impending  events.  With  those  capacity  would  be  possible  to  set  up  Crisis  and  Observatories   for  financial  and  economic,  for  armed  conflicts,  for  crime  and  corruption,  for  social  crisis,  for   health  risks  and  disease  spreading,  for  environmental  changes.   • Global   Development.   An   inspiring   example   is   given   by   Global   Pulse47,   which   is   a   Big   Data   based  innovation  programme  fostered  by  the  UN  Secretary-­‐General  and  aimed  at  harnessing   today's   new   world   of   digital   data   and   real-­‐time   analytics   in   order   foster   international   development,   protect   the   world's   most   vulnerable   populations,   and   strengthen   resilience   to   global   shocks.     The   programme   is   rooted   on   three   main   pillars:   research   on   new   data   indicators   providing   real-­‐time   understanding   of   community’s   welfare   as   well   as   real-­‐time   feedback  on  policies;  creation  of  a  toolkit  of  free  open-­‐source  software  for  mining  real-­‐time   data   useful   for   shared   evidence-­‐based   decisions;   the   establishment   of   country-­‐level                                                                                                                             42     44   45     46   47­‐new   43 67  |  P a g e  
  • 68.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   innovation  centres  (Pulse  Lab)  where  real-­‐time  data  are  applied  to  development  challenges.   The  programme  encompasses  5  main  projects  carried  out  with  several  partners:   o o “Unemployment   through   the   Lens   of   Social   Media” 49 ,   which   relates   the   unemployment   statistics   with   unemployment-­‐related   conversation   from   open   social   web   o “Twitter   and   the   Perception   of   Crisis   Related   Stress” 50,   which   investigates   what   indicators   can   help   in   understanding   people’s   concerns   on   food,   fuel,   finance,   housing     o “Monitoring   Food   Security   Issues   through   New   Media” 51 ,   which   finds   emerging   trends   related   to   food   security   using   text   analysis,   semantic   clustering   and   networks   theory   o • “Daily   Tracking   of   Commodity   Prices:   the   e-­‐Bread   Index”48,   which   investigates   how   scraping  online  prices  could  provide  real-­‐time  insights  on  price  dynamics   “Global   Snapshot   of   Wellbeing   –   Mobile   Survey”52,   aimed   at   experimenting   new   tools  able  of  replicating  the  standards  of  traditional  household  surveys  in  real-­‐time   on  a  global  scale   Intelligence  and  security.  As  examples  of  governments’  commitment  to  Big  Data  for  national   security   we   can   present   the   Cyber-­‐Insider   Threat   (CINDER) 53  program,   which   aims   at   developing   new   ways   for   detecting   cyber   espionage   activities   in   military   computer   networks   as  well  as  at  increasing  the  accuracy,  rate  and  speed  with  which  cyber  threats  are  detected.   Another   example   is   the   Anomaly   Detection   at   Multiple   Scales   (ADAMS)54  program   led   by   the   Defense   Advanced   Research   Project   Agency   (DARPA),   which   addresses   the   problem   of   anomaly-­‐detection   and   characterization   in   massive   data   sets.   The   program   will   be   initially   applied  to  insider-­‐threat  detection,  in  which  individual  actions  are  recognized  as  anomalous   with   comparison   to   a   background   of   routine   network   activity.   Finally   the   Center   of   Excellence   on   Visualization   and   Data   Analytics   (CVADA)   of   the   Department   for   Homeland   Security   (DHS)   is   leading   a   research   effort   on   data   that   can   be   used   by   first   responders   to   tackle   with   natural   disasters   and   terrorists   attacks,   by   law  enforcement   to   border   security   concerns,  or  to  detect  explosives  and  cyber  threats.   An  Interesting  Application:  Smart  Cities   A  Smart  City  is  a  public  administration  or  authorities  delivering  services  and  infrastructure  based  on   ICT  which  are  easy  to  use,  efficient,  responsive,  open  and  sustainable  for  the  environment.  We  can   identify  six  main  dimensions55:     • Smart  economy,   characterized   by   high   standard   of   living   and   competitive   elements:   innovative   and   entrepreneurship,   high   productivity,   flexibility   of   labour   market,   internationalism,  ability  to  transform;   • Smart  mobility,   i.e.   efficient   public   transportation   system,   local   and   international   accessibility,  availability  of  ICT-­‐infrastructure,  sustainability  and  safety;                                                                                                                             48­‐global-­‐prices-­‐local-­‐products-­‐real-­‐time-­‐e-­‐pricing-­‐bread­‐social-­‐media-­‐mining-­‐add-­‐depth-­‐unemployment-­‐statistics   50­‐and-­‐perceptions-­‐crisis-­‐related-­‐stress   51­‐awareness-­‐and-­‐emergent-­‐information-­‐monitoring-­‐system-­‐food-­‐security   52­‐snapshot-­‐wellbeing-­‐mobile-­‐survey   53­‐Insider_Threat_%28CINDER%29.aspx   54   55  See  also  the  project  EuropeanSmartCities  at­‐ 49 68  |  P a g e  
  • 69.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Smart  environment   (sustainability   of   natural   resources):   low   pollution,   protection   of   environment,  natural  attractiveness;     • Smart  people,   given   by   high   level   of   human   and   intellectual   capital,   high   level   of   qualification,  lifelong  learning,  social  and  ethnic  diversity,  flexibility,  creativity;     • Smart  living   (high   quality   of   life);   presence   of   cultural   facilities,   healthy   environment   conditions,  individual  safety,  housing  quality,  education  facilities,  touristic  attractiveness  and   social  cohesion;   • Smart  governance  given  by  citizens’  participation  in  decision-­‐making,  the  presence  of  public   and  social  services  and  of  transparent  and  open  governance.   The   combination   of   all   the   benefits   stemming   from   Big   Data   in   governance,   make   it   evident   that   the   integration  of  heterogeneous  data  from  various  domains  holds  high  potential  to  provide  insights  on   cities.  New  technologies  will  unlock  massive  amounts  of  data  about  all  the  aspects  of  the  city  as  well   as  its  citizens.  For  instance  new  systems  involving  energy  use  at  fixed  locations  (point  sources,  like   house  and  office)  are  being  implemented  by  the  mean  of  smart  metering  as  well  as  the  integration   of  various  information  systems  used  to  record  pricing  and  activity.  Another  possibility  is  given  by  the   extraction  of  positional  and  frequency  data  from  social  media  such  as  Twitter,  Facebook,  Flickr  and   Foursquare.  All  this  data  will  be  used  for  fulfilling  the  Smart  Cities  targets.  Let  us  take  into  account   for  instance  the  transportation  system,  where  diagnosing  and  anticipating  abnormal  events  such  as   traffic   congestions   requires   integration   of   various   data   like   traffic   data,   weather   data,   road   conditions,  or  traffic  light  strategy.  Another  possibility  will  be  given  by  e-­‐inclusion  technologies  and   open   data   for   governance.   One   important   example   of   the   development   of   the   Smart   City   concept   at   large   scale   is   the   New   York   City   project   “Roadmap   for   a   Digital   Future”56,   which   outlines   a   path   to   build  on  New  York  City's  successes  and  establish  it  as  the  world's  top-­‐ranked  Digital  City,  based  on   indices  of  internet  access,  open  government,  citizen  engagement,  and  digital  industry  growth.     Recent  Trends   Big   Data   is   a   fast   growing   phenomenon:   as   the   Google   CEO   Eric   Schmidt   pointed   out   in   2010,   currently  in  two  days  is  created  in  the  world  as  much  information  as  it  was  from  the  appearance  of   man   till   2003.   Nowadays57  it   is   possible   to   store   all   the   world’s   music   in   a   $600   worth   disk   drive,   while  Facebook  content  shared  every  month  amounts  to  $30  billion.  According  to  the  forecast  global   data  will  grow  at  a  40%  rate  next  year  while  the  total  IT  spending  will  grow  just  by  5%.   In  2010  users   and  companies  stored  more  than  13  exabytes  of  new  data,  which  is  over  50,000  times  the  data  in   the  Library  of  Congress.     Big  Data  is  also  a  potential  booster  for  the  economy,  bearing  a  $300  billion  potential  annual  value  to   US   health   care   as   well   as   a     $600   billion   potential   annual   consumers   surplus   from   using   personal   location   data   globally   and   a   250   billion   Euro   potential   annual   value   to   European   public   administration.  In  fact  the  European  Commission  is  expected  to  adopt  an  Open  Data  Strategy,  i.e.  a   set   of   measures   aimed   at   increasing   government   transparency   and   creating   a   €32   billion   a   year   market   for   public   data.   Finally   as   reported   last   year   by   the   McKinsey   Global   Institute58,   the   United   States   will   need   140,000   to   190,000   more   workers   with   deep   analytical   expertise   and   1.5   million   more  data-­‐literate  managers.  Always  according  to  the  McKinsey  Global  Institute  the  potential  value   of  global  personal  location  data  is  estimated  to  be  $700  billion  to  end  users,  and  it  can  result  in  an  up   to  50%  decrease  in  product  development  and  assembly  costs.  What’s  the  growth  engine  of  big  data?                                                                                                                             56    See  McKinsey  Global  Institute  (2011)  “Big  data:  The  next  frontier  for  innovation,  competition,  and  productivity”     58     57 69  |  P a g e  
  • 70.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   From  one  side  more  “old  world”  data  is  produced  through  “open  governance”  and  digitization.  From   the  other  side  “new  world”  data  are  created  are  continuously  collected  in  domains  such  as  “in  silico”   medicine,  “in  silico  engineering”  and   Internet  science.  Brand  new  fields  of  science  are  being  created:   computational   chemistry,   biology,   economics,   engineering,   mechanics,   neuroscience,   geophysics,   etc.  etc.  This  is  true  also  in  humanities,  such  as  the  birth  of  computational  social  science,  based  on   mobile   phones   and   social   network   digital   traces.   A   wide   array   of   actors   including   humanities   and   social   science   academics,   marketers,   governmental   organizations,   educational   institutions,   and   motivated   individuals,   are   now   engaged   in   producing,   sharing,   interacting   with,   and   organizing   data.   All  these  developments  are  allowed  by  the  rise  of  new  technologies  for  data  collections:  web  logs;   RFID;  sensor  networks;  social  networks;  social  data  (due  to  the  Social  data  revolution),  Internet  text   and   documents;   Internet   search   indexing;   call   detail   records;   astronomy,   atmospheric   science,   genomics,   biogeochemical,   biological;   military   surveillance;   medical   records;   photography   archives;   video  archives;  large-­‐scale  eCommerce.     Inspiring  cases   • The   Ion   ProtonTM   Sequencer59  is   a   rapid   genome-­‐scale   benchtop   sequencer.   The   tool   allows  to  perform  data  analysis    in  the  same  day  on  a  single  stand-­‐alone  server.   • The   NIH   Human   Connectome   Project 60  aims   at   mapping   the   neural   pathways   that   underlie  human  brain  function  in  order  to  acquire  and  share  data  about  the  structural   and  functional  connectivity  of  the  human  brain.   • The   Models   of   Infectious   Disease   Agent   Study 61  is   a   collaboration   of   research   and   informatics   groups   to   develop   computational   models   of   the   interactions   between   infectious   agents   and   their   hosts,   disease   spread,   prediction   systems   and   response   strategies.                                                                                                                 • 62  is   a   portal,   developed   by   the   Land   Transport   Authority   (LTA)   of   Singapore,  providing  information  and  eServices  for  all  land  transport  users.   • UN   Global   Pulse63,   an   innovation   initiative   launched   by   the   United   Nations   Secretary-­‐ General   aimed   at   exploring   how   digital   data   sources   and   real-­‐time   analytics   technologies  can  help  policymakers  to  better  protect  populations  from  shocks.   Tools  on  the  market   Freely  available  tools   There  are  not  many  cases  of  freely  available  tools  for  Big  Data  analysis  on  the  market.   The  presence  of  freely  available  tools  on  the  market  bear  many  benefits,  such  as:   • Developers  and  analysts  will  use  them  to  experiment  with  emerging  types  of  data  structure   so  as  to  develop  new  and  different  analytical  procedures,  he  added                                                                                                                             59­‐us/news-­‐gallery/press-­‐releases/2012/life-­‐techologies-­‐ itroduces-­‐the-­‐bechtop-­‐io-­‐proto.html.html   60     61     62     63­‐new   70  |  P a g e  
  • 71.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Developers   and   IT   professionals   contribute   their   findings   and   know-­‐how   back   into   the   industry  to  drive  knowledge  exchange   The   freely   available   tools   permit   to   overcome   data   limitations,   simplify   the   analytical   process   and   visualize  results.  The  functionalities  provided  by  these  software  are:   • Massively  parallel  processing  (MPP)  database  product  for  large-­‐scale  analytics  and  next-­‐gen   data  warehousing   • Data-­‐parallel  implementations  of  statistical  and  machine  learning  methods   • Visual  data  mining  modelling     In  this  view  are  very  important  the  free  Big  Data  tools  developed  by  Greenplum  for  data  scientists   and  developers:  MADlib  and  Alpine  In-­‐Database  Miner64  and  Greenplum  HD  Community  Edition65.   Some   other   software   partially   for   free   with   important   Big   Data   applications:   KNIME 66,   Weka   /   Pentaho67,   Rapid-­‐I   RapidAnalytics68,   Rapid-­‐I   RapidMiner69.   Finally   there   is   R70,   which   although   was   not  built  for  Big  Data,  it  has  interesting  application  in  this  realm.       Enterprise-­‐level  software   The  enterprise-­‐level  software  is  adopted  for  the  following  functionalities:   • Open  source  software  based  on  Apache  Hadoop     • Data  storage  platforms  and  other  information  infrastructure  solutions   • Shared-­‐nothing  massively  parallel  processing  (MPP)  database  architectures   • Dataflow  engines,  software  interconnect  technologies   • Data  discovery  and  exploration  tools   • Built-­‐in  text  analytics,  enterprise-­‐grade  security  and  administrative  tools   • Real-­‐time  analytic  processing  (RTAP)  platforms   • Software-­‐as-­‐a-­‐service  (SaaS)   • Visualization  features  supporting  exploratory  and  discovery  analytics   • On-­‐line  analytical  processing  (OLAP)   • BI/DW  (business  intelligence  and  data  warehousing)   • EDW  (enterprise  data  warehousing).  Examples  of  these  software  include:  Tableau  BI  platform71;   SAS   Data   Integration   Studio 72;   SAS   High   Performance   Analytics 73 ;   SAS   On   Demand 74 ;   SAND                                                                                                                             64­‐tools/­‐ce/   66     67   68  http://rapid-­‐     69  http://rapid-­‐   70  http://www.r-­‐     71     72     73­‐performance-­‐analytics/in-­‐memory-­‐analytics/analytics.html     74     65 71  |  P a g e  
  • 72.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   Analytic   Platform 75;   SAP   BEx 76;   SAP   NetWeaver 77 ;   SAP   In-­‐Memory   Appliance   (SAP   HANA) 78 ;     ParAccel   Analytic   Database   (PADB) 79 ;   IBM   Netezza 80 ;   IBM   InfoSphere   BigInsights 81 ;   IBM   InfoSphere  Streams82;  Kognitio  WX283;  Kognitio  Pablo84;  EMC  Greenplum  Database85;  Greenplum   HD86;  EMC  Greenplum  Data  Computing  Appliance87;  Greenplum  Chorus88;  Cloudera  Enterprise89,   StatSoft  Statistica90  .   Some   other   software   which   have   not   been   built   specifically   for   Big   Data   applications,   but   nonetheless  can  be  used  for  Big  Data  analytics  are:  Mathematica91,  MatLab92  and  Stata93.     Key  challenges  and  gaps   In   order   to   enjoy   all   the   potential   stemming   from   Big   Data   it   would   be   necessary   to   remove   the   technological   barrier   preventing   the   exchange   of   data,   information   and   knowledge   between,   disciplines,   as   well   as   to   integrate   activities   which   are   based   on   different   ontological   foundations.   Even  though  Big  Data  have  provided  a  lot  of  benefits,  many  challenges  are  still  to  be  coped  with.  For   instance  Gartner  (2011)94  argues  that  the  challenges  are  not  only  given  by  the  volume  of  data,  but   also   by   the   variety   (heterogeneity   of   data   types   and   representation,   semantic   interpretation)   and   velocity   (rate   of   data   arrival   and   action   timing).   According   to   the   recent   research   those   advancements  include95   • Data   modelling   challenges:   data   models   coherent   to   the   data   representation   needs;   data   models   able   to   describe   discipline   specific   aspects;   data   models   for   representation   and   query   of   data   provenance   and   contextual   information;   data   models   and   query   languages   representing   and   managing   data   uncertainty,   and   representing   and   querying   data   quality   information   • Data   management   challenges:   provide   quality,   cost-­‐effective,   reliable   preservation   and   access   to   the   data;   protect   property   rights,   privacy   and   security   of   sensible   data;   ensure   data  search  and  discovery  across  a  wide  variety  of  sources;  connect  data  sets  from  different   domains   in   order   to   create   open   linked   data   space   data   can   be   unstructured   or   semi-­‐ structured  with  no  context;  different  data  format;  different  data  labels  used  for  same  data   elements;  different  data  entry  conventions  and  vocabularies  used;  -­‐  data  entry  errors;  data                                                                                                                             75­‐explorer     77     78­‐memory-­‐computing-­‐platform/hana/overview/index.epx   79   80  http://www-­‐   81  http://www-­‐   82  http://www-­‐   83   84   85­‐database   86­‐hd   87­‐dca   88   89­‐services/enterprise/   90     91     92   93     94   2011.    Available  at   95­‐b21d-­‐4b90-­‐81d4-­‐d909fdb96b87.pdf   76 72  |  P a g e  
  • 73.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   sets   can   be   so   large   they   cannot   be   effectively   processed   by   a   single   machine;     data   parallelization  and  task  parallelization96.   Data   service/tools   challenges:   data   tools   for   most   scientific   disciplines   are   inadequate   to   support  research  in  all  its  phases  so  that  scientists  are  less  productive  than  what  they  might   be.  In  fact  there  is  the  need  of  software  able  to  “clean”,  analyse  and  visualize  huge  amounts   of   data.   Moreover   are   missing   data   tools   and   policies   for   the   ensuring   the   cross   collaboration  and  fertilization  among  different  disciplines  and  scientific  realms   •   As  for  other  issues  concerning  Big  Data,  Boyd  and  Crawford  (2011)  highlight  some  of  them:   • Relationship  between  automatic  search  and  the  definition  of  knowledge.   At   the   beginning   of   the   20th   century   Ford   introduced   the   mass   production,   automation   and   assembly   line,   reshaping  not  only  the  way  things  are  produced,  but  also  the  general  understanding  of  labor,   the   human   relationship   to   work,   and   the   society   at   large.   Fordism   consisted   in   breaking   down  holistic  tasks  into  atomized  and  independent  ones.  In  the  same  way  Big  Data  is  a  new   system  of  knowledge  characterized  by  a  computational  turn  in  science  leading  to  a  change  in   the   constitution   of   knowledge,   the   process   of   research   and   the   categorization   of   reality.   But   as  the  Fordism  had  limits  (indeed  has  been  overcome  by  the  Just  in  Time  paradigm),  also  the   specialized   Big   Data   tools   are   not   flawless.   Big   Data,   as   a   new   system   of   knowledge   can   change   the   very   meaning   of   learning   itself,   with   all   the   possibilities   and   limitations   embedded  in  the  systems  of  knowing     • Big  Data  may  produce  misleading  claims  of  objectivity  and  accuracy.  In  the  science  there  is   a   deep   cleavage   between   qualitative   and   quantitative   scientists.   Apparently   qualitative   scientists  would  be  engaged  in  creating  and  interpreting  stories,  while  quantitative  scientists   would   be   in   the   business   of   producing   facts.   Needless   to   say,   that   is   not   case   as   all   the   objectivity   claims   come   from   subjects,   who   make   subjective   observation   and   choices.   Moreover   data   analysis   is   based   on   a   large   number   of   assumptions   (see   for   instance   the   asymptotic   theory   in   statistics)   and   on   the   other   hand   even   though   a   model   may   be   mathematically   or   an   experiment   may   be   scientifically   valid,   the   final   interpretation   is   subjective.   Other   examples   are   the   difficulty   of   integrating   in   a   consistent   way   different   datasets,   the   arbitrary   choices   inherent   data   cleaning   and   finally   the   fact   that   internet   databases  may  well  be  affected  by  bias  such  as  frictions  and  self-­‐selection.  In  this  view,  by   increasing   the   quantification   space,   especially   in   social   sciences,   Big   Data   might   support   objectivity  and  accuracy  claims  which  are  not  really  grounded  on  good  sense  and  reality.       • A   higher   quantity   of   data   does   not   always   mean   better   data.   In   all   sciences   there   is   a   massive   amount   of   literature   (interpretation   bias,   design   standardization,   sampling   mechanism  and  question  bias,  statistical  significance  and  diagnostics)  aimed  at  ensuring  the   consistency  of  data  collection  and  analysis.  Curiously,  Big  Data  scientists  sometimes  assume   a   priori   quality   of   their   data   and   completely   neglects   the   methodological   issues   proper   of   global   sciences.   A   clear   example   is   given   by   social   media   data,   which   are   subject   to   self-­‐                                                                                                                           96  Some  Big  Data  challenges  are  deeply  related  with  policy  making,  such  as  the  fact  that  many  agencies  pay  a  high  premium   to   both   internal  resources  and  external  third  parties  to  manage  their  data.  Additionally,  data  management  can  sometimes   be   redundant   if   not   properly   set   up.   Moreover   regulations   do   not   take   into   account   the   new,   expanded   capabilities   that   IT   offers  as  it  takes  time  to  issue  a  new  law  and  bureaucrats  are  not  so  keen  to  novelty     73  |  P a g e  
  • 74.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   selection   bias   as   people   using   social   media   is   not   representative   of   the   society   itself.   Even   the   definition   of   active   user   and   account   of   a   social   media   might   not   be   innocuous:   in   fact   it   is  estimated  that  40%  of  Twitter’s  users  are  merely  “listeners”,  i.e.  do  not  proactively  take   part.   Finally   it   has   to   be   recognized   that   in   my   contexts   high   quality   research   is   purposely   carried   out   with   a   limited   amount   of   data,   such   as   for   instance   in   game   theory   experimental   analysis.     • Big  Data  and  Ethical  Issues.  The  use  for  research  purposes  of  “public”  data  on  social  media   websites   opens   the   door   to   deontological   issues.   The   problem   is:   can   those   data   be   used   without   any   ethical   of   privacy   consideration?   Obviously   Big   Data   is   an   emerging   field   of   science,   thereby   ethical   consideration   are   yet   to   be   fully   considered.   How   the   researchers   can   be   sure   that   their   activity   is   not   harmful   for   some   of   their   subjects?   On   one   hand   is   impossible  to  ask  for  data  use  permission  from  all  the  subjects  present  in  a  database.  On  the   other   hand,   the   mere   fact   that   the   data   are   available   does   not   justify   their   use.   Accountability   to   the   field   of   research   and   accountability   to   the   research   subjects   are   the   ethical   keys   for   Big   Data.   In   all   the   traditional   fields   of   science,   researcher   must   follow   a   series   of   professional   standards   aimed   at   protecting   the   rights   and   well   being   of   human   subjects.  On  the  other  hand  the  ethical  implications  of  Big  Data  research  are  not  yet  clear.     • Digital   divides   created   by   Big   Data.   It   is   widely   accepted   that   doing   research   on   Big   Data   automatically  involves  having  a  quick  and  easy  access  to  databases.  This  is  not  the  case,  as   only   social   media   companies   have   access   to   large   datasets,   and   sell   those   data   at   a   high   price,   offering   only   small   data   sets   to   university   based   researchers.   So   researchers   with   a   considerable   amount   of   founding   or   based   inside   those   firms   can   have   access   to   data   that   the   outsiders   will   not.   Thereby   their   methodologies   and   claims   cannot   be   verified.   In   this   view   Big   Data   can   create   a   new   digital   divide,   between   researchers   belonging   to   the   top   universities   and   working   with   the   top   companies,   and   scholar   belonging   to   the   periphery.   But  the  digital  divide  can  be  also  skills  based:  in  fact  only  people  with  a  strong  computational   background   are   able   to   wrangle   through   APIs   and   analyse   massive   quantities   of   data.   Concluding  there  is  a  new  digital  divide  between  the  Big  Data  reach,  who  are  able  to  analyze   and  to  buy  datasets,  and  belong  to  top  universities  and  companies,  and  the  Big  Data  poor,   who  are  outsiders     Finally  according  to  the  UN97  the  Big  Data  challenges  can  be  divided  along  two  main  dimensions.   The  data  management:   • Privacy.   The   development   of   new   technologies   always   raises   privacy   concerns   for   individuals,   companies   and   societies.   This   is   a   very   crucial   issue   as   privacy,   safety   and   diversity  are  important  for  defending  the  freedom  of  citizens,  and  obviously  companies  have   the   right   to   retain   their   confidential   information.   In   the   era   of   Big   Data,   the   primary   producers,   who   are   the   citizens   using   services   and   devices   generating   data,   are   seldom   aware  that  they  are  doing  so  or  how  their  data  will  be  used.  Sometimes  it  is  also  unclear  to   what  extent  users  of  social  media  such  as  Twitter  consent  to  the  analysis  of  their  data.  The   pool   of   individual   information   shared   by   mobile   phones   and   credit   card   companies,   social                                                                                                                             97   74  |  P a g e  
  • 75.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   media   and   research   engines   is   simply   astonishing.   People   must   be   conscious   of   that,   as   privacy  is  a  freedom  pillar.         • Access  and  sharing.  A  great  amount  of  data  is  available  online  for  the  most  disparate  uses.   On   the   other   hand   much   data   is   retained   by   companies   which   are   concerned   about   their   reputation,  the  necessity  to  protect  their  competitiveness  or  simply  lack  the  right  incentive   to  do  so.  On  the  other  hand  there  is  a  bunch  of  technical  and  regulatory  arrangements  which   has  to  be  put  in  place  in  order  to  ensure  inter-­‐comparability  of  data  and  interoperability  of   systems.       Data  analysis:   • Summarising   the   data.   Sometimes   the   data   might   be   simply   false   or   fabricated,   especially   with   user-­‐generated   text-­‐based   data   (blogs,   news,   social   media   messages).   In   addition   sometimes   data   are   derived   from   people’s   perceptions,   as   in   calls   to   health   hotlines   and   online   searches   for   symptoms.   Another   case   is   related   to   opinion   mining   and   sentiment   analysis,  in  which  the  true  significance  of  the  statements  can  be  misled,  so  that  the  human   factor   is   always   crucial   in   the   analysis.   Another   problem   is   that   sometimes   data   are   generated   from   expressed   intentions   in   blogposts,   online   searches,   mobile-­‐phone   systems   for   checking   market   price,   which   are   not   a   sure   indicator   of   actual   intentions   and   final   decisions.   So   there   is   a   huge   problem   in   summarizing   facts   from   users’   generated   text,   as   there  might  be  a  difficulty  in  distinguishing  feeling  from  facts.   • Interpreting   data.  A  very  important  concern  is  given  by  the  sample  selection  bias,  given  by   the   fact   that   people   generating   data   are   not   representative   of   the   entire   population.   For   instance   younger   generations   use   more   internet   and   mobile   devices.   In   this   way   the   conclusions   of   the   analysis   are   valid   only   for   the   sample   at   hand   and   cannot   therefore   be   generalized.  Sometimes  dealing  with  huge  amounts  of  data  leads  the  researchers  to  focus  on   finding   patterns   or   correlations   without   concentrating   on   the   underlying   dynamics.   One   thing   is   to   find   a   correlation,   another   is   to   detect   a   causal   relationship.   Even   more   difficult   is   to   identify   the   direction   of   the   causal   relationship   without   using   a   founding   theory.   A   final   issue   is   very   much   linked   with   using   data   from   different   sources,   which   can   magnify   the   existing  flaws  in  each  database     Finally   we   have   the   challenges   identified   by   the   community   white   paper   drafted   with   the   collaboration  of  a  group  of  leading  researchers  across  the  United  States98:   • Heterogeneity   and   incompleteness.   Data   must   be   structured   prior   to   the   analysis   in   an   homogeneous   way,   as   algorithms   unlike   humans   are   not   able   to   grasp   nuance.   Most   computer   systems   work   better   if   multiple   items   are   stored   in   an   identical   size   and   structure.   But   an   efficient   representation,   access   and   analysis   of   semi-­‐structured   data   is   necessary   because   as   a   less   structured   design   is   more   useful   for   certain   analysis   and   purposes.   Even   after   cleaning   and   error   correction   in   the   database,   some   errors   and   incompleteness   will   remain,  challenging  the  precision  of  the  analysis.   • Human   collaboration.   Even   if   analytical   instruments   gained   tremendous   advancements,   there  are  still  many  realms  in  which  the  human  factor  is  able  to  discover  patterns  algorithms                                                                                                                             98   75  |  P a g e  
  • 76.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   cannot.  An  example  can  be  found  in  the  use  of  CAPTCHAs,  which  can  discern  human  users   from   computer   programmes.   In   this   view   a   Big   Data   system   cannot   must   involve   a   human   presence.   Given   the   complexity   of   today’s   world,   there   is   the   necessity   to   harness   human   ingenuity   from   different   domains   through   crowdsourcing.   Thereby   a   Big   Data   system   requires   technologies   able   to   support   this   kind   of   collaboration   even   in   case   of   conflicting   statements  and  judgments.       Current  Big  Data  Techniques     Big   datasets   can   be   analysed   by   the   mean   of   several   techniques   coming   from   statistics   and   computers  science.     A  list  of  the  principal  categories  is:   • Cluster   analysis.   Statistical   technique   consisting   in   splitting   an   heterogeneous   group   into   smaller   subsets   of   similar   elements,   whose   characteristics   of   similarities   are   not   known   in   advance.  A  typical  example  is  to  identify  consumers  with  similar  patterns  of  past  purchases   in  order  to  tailor  most  accurately  a  given  marketing  strategy   • Crowdsourcing.   Technique   for   the   collection   of   data   which   have   been   drawn   from   a   large   group   or   community   in   response   to   an   open   call   through   a   networked   media   such   as   the   internet.   This   category   bears   a   crucial   importance   in   our   case   as   it   is   a   mass   collaboration   instance  of  using  Web  2.0   • Data   mining.   Combination   of   database   management,   statistics   and   machine   learning   methods  useful  for  extracting  patterns  from  large  datasets.  Some  examples  include  mining   human  resources  data  in  order  to  assess  some  employee  characteristics  or  consumer  bundle   analysis  to  model  the  behavior  of  customers   • Machine   learning.   Subfield   of   computer   science   (in   the   scope   of   artificial   intelligence)   regarding  the  definition  and  the  implementation  of  algorithms  allowing  computers  to  evolve   their  behaviour  based  on  empirical  evidence.  An  example  of  machine  learning  is  the  natural   language  processing.   • Natural   language   processing.   Set   of   computer   science   and   linguistic   methods   adopting   algorithms  to  analyse  natural  human  language.  Basically  this  field,  which  began  as  a  branch   of  artificial  intelligence,  deals  with  the  interaction  between  computer  and  human  language   • Neural   networks.   Computational   models   which   are   structured   and   work   similarly   to   biological  neural  networks  existing  among  brain  cells,  and  that  are  used  to  find  in  particular   non-­‐linear  patterns  in  the  data.  Some  applications  include  game-­‐playing  and  decision  making   (backgammon,  chess,  poker)  and  knowledge  discovery  in  data  bases     • Network   analysis.   Part   of   graph   theory   and   network   science   which   describes   the   relationships   among   discrete   nodes   in   a   graph   or   a   network.   In   particular   the   social   network   analysis   studies   the   structure   of   relationship   among   social   entities.   Some   applications   are   include   the   role   of   trust   in   exchange   relationships   and   the   study   of   recruitment   into   political   movements  and  social  organizations   • Predictive   modelling.  Branch  a  mathematical  model  used  to  best  predict  the  probability  of   an  outcome.  This  technique  is  widely  used  in  customer  relationship  management  to  produce   customer-­‐level  models  able  to  assess  the  probability  that  a  customer  would  take  a  particular   action,  such  as  cross-­‐sell,  product  deep-­‐sell  and  churn   76  |  P a g e  
  • 77.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Regression.  Statistical  method  for  assessing  how  the  value  of  a  dependent  variable  changes   with   one   or   more   dependent   variables.   Examples   of   applications   include   the   change   in   consumer’s  behaviour  due  to  manufacturing  parameters  or  economic  fundamentals   • Sentiment  analysis.  Natural  language  processing  methods  for  extracting  information  such  as   polarity,  degree  and  strength  of  the  sentiment  over  a  given  feature,  aspect  of  product.  Many   companies   assess   how   different   customers   and   stakeholders   react   to   their   products   and   action  by  applying  this  analysis  to  blogs,  social  networks  and  other  social  media   • Spatial   analysis.   Methods   for   assessing   the   geographical,   geometric   or   topological   characteristics  of  a  data  set.  The  spatial  data  are  often  drawn  from  geographical  information   systems  (GIS)  including  addresses  or  latitude/longitude  coordinates,  to  be  incorporated  into   spatial  regressions  (correlation  between  commodity  price  and  location)  or  simulations   • Simulation.  Consists  in  modelling  the  behavior  of  a  complex  system  for  performing  forecast   and   scenario   analysis.   As   example   we   can   mention   Monte   Carlo   simulations,   which   are   a   class  of  computational  algorithms  that  rely  on  repeated  random  sampling  to  compute  their   results     Current  and  Future  Research   • Technologies   for   collecting   cleaning,   storing   and   managing   data:   datawarehouse;   pivotal   transformation;   ETL;   I/O;   efficient   archiving,   storing,   indexing,   retrieving,   and   recovery;   streaming,   filtering,   compressed   sensing   sufficient   statistics;   automatic   data   annotation;   Large   Database   Management   Systems;   storage   architectures;   data   validity,   integrity,   consistency,   uncertainty   management;   languages,   tools,   methodologies   and   programming   environments   • Technologies   for   summarizing   data   and   extracting   some   meaning:   reports;   dashboard;   statistical   analysis   and   inference;   Bayesian   techniques;   information   extraction   from   unstructured,   multimodal   data;   scalable   and   interactive   data   visualization;   extraction   and   integration   of   knowledge   from   massive,   complex,   multi-­‐modal,   or   dynamic   data;   data   mining;   scalable   machine   learning;   data-­‐driven   high   fidelity   simulations;   scalable   machine   learning;   predictive   modelling,   hypothesis   generation   and   automated   discovery   Technologies   for   using   data   a   decision   tool:   Decision   Trees,   Pro-­‐Con   Analysis,   Rule   Based   Systems,   Neural   Networks,   Tradeoff   based   Decisions   (which   incorporates   Reporting,   Statistics,  Knowledge  Based  systems)     77  |  P a g e  
  • 78.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     3.2.2. Opinion  Mining  and  Sentiment  Analysis   Summary  overview     Current  free  tools   Top   market   tools   -­‐  Filtering  opinion  based   on   rating;   assessing   sentiments   based   on   keywords;   visual   word   counting   -­‐    Machine  ·∙    -­‐        -­‐  Statistical  +  Semantic  analysis   ·∙              -­‐  Visual   ·∙                -­‐  Multilingual   learning  +   through  lexicon/corpus  of  words   representation   audiovisual   human   with  known  sentiment  for   opinion  mining   ·∙                -­‐  Audiovisual   analysis   sentiment  classification   opinion  mining   ·∙                -­‐  Usable,  peer-­‐to-­‐ ·∙                -­‐  Identification  of  policy  -­‐   peer  opinion   ·∙                -­‐  Real-­‐time  opinion   opinionated  material  to  be   mining  tools  for   mining   analysed   citizens            ·∙      -­‐  Machine  learning   ·∙                -­‐  Computer-­‐generated  reference   ·∙                -­‐  Non-­‐bipolar   algorithms   corpuses  in  political/governance   assessment  of   field   opinion            ·∙    -­‐  Natural  language   interfaces   ·∙                  -­‐  Visual  mapping  of  bipolar   ·∙                -­‐  Automatic  irony   opinion   detection   ·∙                -­‐  SNA  applied  to   opinion  and   ·∙                  -­‐  Identification  of  highly  rated     expertise   experts   ·∙                -­‐  Bipolar  assessment   of  opinions   -­‐  Argument  mapping   Current  research   Short   term   Long   term   future  research   future   research   ·∙                -­‐  Multilingual   reference  corpora   ·∙                -­‐  Recommendation   algorithms     Introduction  and  definition   The   explosion   of   social   media   has   created   unprecedented   opportunities   for   citizens   to   publicly   voice   their  opinions,  but  has  created  serious  bottlenecks  when  it  comes  to  making  sense  of  these  opinions.   At   the   same   time,   the   urgency   to   gain   a   real-­‐time   understanding   of   citizens   concerns   has   grown:   because   of   the   viral   nature   of   social   media   (where   attention   is   very   unevenly   distributed)   some   issues  rapidly  and  unpredictably  become  important  through  word-­‐of-­‐mouth.   Policy-­‐makers  and  citizens  don’t  yet  have  an  effective  way  to  make  sense  of  this  mass  conversation   and  interact  meaningfully  with  thousands  of  others.   As  a  result  of  this  paradox,  the  public  debate  in  social  media  is  characterized  by  short-­‐termism  and   auto-­‐referentiality.   Many   experts   consider   social   media   as   a   missed   opportunity   for   better   policy   debate.   At   the   same   time,   the   sheer   amount   of   raw   data   is   also   an   opportunity   to   better   make   sense   of   opinions.   The   key   asset   that   Google   exploited   to   reach   dominance   in   the   search   market   is   not   a   better  algorithm,  but  the  power  of  more  data.   78  |  P a g e  
  • 79.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   We  are  therefore  at  a  crucial  underpinning  where  the  challenge  of  information  overload  can  become   not  a  problem,  but  an  opportunity  for  making  sense  of  a  thousand  voices  and  identify  problems  as   soon  as  they  arise.     Opinion   mining   can   be   defined   as   a   sub-­‐discipline   of   computational   linguistics   that   focuses   on   extracting   people’s   opinion   from   the   web.   The   recent   expansion   of   the   web   encourages   users   to   contribute   and   express   themselves   via   blogs,   videos,   social   networking   sites,   etc.   All   these   platforms   provide  a  huge  amount  of  valuable  information  that  we  are  interested  to  analyse.  Given  a  piece  of   text,  opinion-­‐mining  systems  analyse:   ·∙              Which  part  is  opinion  expressing;   ·∙              Who  wrote  the  opinion;   ·∙              What  is  being  commented.   Sentiment   analysis,   on   the   other   hand,   is   about   determining   the   subjectivity,   polarity   (positive   or   negative)  and  polarity  strength  (weakly  positive,  mildly  positive,  strongly  positive,  etc.)  of  a  piece  of   text  –  in  other  words:   ·∙              What  is  the  opinion  of  the  writer       Opinion  mining  and  sentiment  analysis  cover  a  wide  range  of  applications.     1. Argument  mapping  software  helps  organising  in  a  logical  way  these  policy  statements,  by   making  explicit  the  logical  links  between  them.  Under  the  research  field  of  Online   Deliberation,  tools  like  Compendium,  Debatepedia,  Cohere,  Debategraph  have  been   developed  to  give  a  logical  structure  to  a  number  of  policy  statement,  and  to  link  arguments   with  the  evidence  to  back  it  up99.     2. Voting  Advise  Applications  help  voters  understanding  which  political  party  (or  other  voters)   have  closer  positions  to  theirs.  For  instance,  asks  the  voter  to  declare  its   degree  of  agreement  with  a  number  of  policy  statements,  then  matches  its  position  with  the   political  parties.   3. Automated  content  analysis  helps  processing  large  amount  of  qualitative  data.  There  are   today  on  the  market  many  tools  that  combine  statistical  algorithm  with  semantics  and   ontologies,  as  well  as  machine  learning  with  human  supervision.  These  solutions  are  able  to   identify  relevant  comments  and  assign  positive  or  negative  connotations  to  it  (the  so-­‐called   sentiment).     The   first   two   point   reflect   mature   application   areas,   while   the   third   area   is   emerging   and   with   relevant  research  issues.  We  will  therefore  mainly  focus  on  this  area  for  the  research  issues.     Why  it  matters  in  governance   These  applications  are  the  basic  infrastructure  of  large  scale  collaborative  policy-­‐making.  They  help   making   sense   of   thousands   of   interventions.   They   help   to   detect   early   warning   system   of   possible   disruption   in   a   timely   manner,   by   detecting   early   feedback   from   citizens.   Traditionally,   ad   hoc                                                                                                                             99  Other  similar  tools  include  Rationale  (   79  |  P a g e  
  • 80.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   surveys   are   used   to   collect   feedback   in   a   structured   manner.   However,   this   kind   of   data   collection   is   expensive,  as  it  deserves  an  investment  in  design  and  data  collection;  it  is  difficult,  as  people  are  not   interested   in   answering   surveys;   and   ultimately   it   is   not   very   valuable,   as   it   detects   “known   problems”  through  pre-­‐defined  questions  and  interviewees,  but  fails  to  detect  the  most  important   problems,   the   famous   “unknown   unknown”.   Opinion   mining   is   helpful   to   identify   problems   by   listening,  rather  than  by  asking,  thereby  ensuring  a  more  accurate  reflection  of  reality.     Argument  mapping  software  is  then  useful  to  ensure  that  policy  debates  are  logical  and  evidence-­‐ based,  and  do  not  repeat  the  same  arguments  again  and  again.   These   tools   would   finally   be   helpful   not   only   for   policy-­‐makers,   but   also   for   citizens   who   could   more   easily  understand  the  key  points  of  a  discussion  and  participate  to  the  policy-­‐making  process.     Recent  trends   Opinion   mining   is   not   in   itself   a   new   research   theme.   Automated   methods   for   content   analysis   have   been   increasingly   used,   and   have   increased   at   least   6   folds   from   1980   to   2002   (Neuendorf,   K.   A.   2002.   The   Content   Analysis   Guidebook.   Sage).   The   research   theme   is   based   in   long   established   computer   science   disciplines,   such   as   Natural   Language   Processing,   Text   Mining,   Machine   Learning   and  Artificial  Intelligence,  Automated  Content  Analysis,  and  Voting  Advise  Applications.   However,  according  to  Pang  and  Lee  (2008),  since  2001  we  see  a  growing  awareness  of  the  problems   and   opportunities,   and   “subsequently   there   have   been   literally   hundreds   of   papers   published   on   the   subject.”   What   is   new   today   is   the   sheer   increase   in   the   quantity   of   unstructured   data,   mainly   due   to   the   adoption  of  social  media,  that  are  available  for  machine  learning  algorithm  to  be  trained  on.  Social   media   content   by   nature   reflects   opinions   and   sentiments,   while   traditional   content   analysis   tended   to   focus   on   identifying   topics   ((Pang,   Lee,   and   Vaithyanathan   2002).   As   such,   it   deals   with   more   complex  natural  language  problems.  Because  of  the  combination  of  increase  in  the  volume  of  data   available   and   more   complex   concepts   to   analyse,   in   recent   years   there   has   been   a   decrease   in   interest   on   semantic-­‐based   application,   and   a   move   towards   greater   use   of   statistics   and   visualisation.   Just   as   any   other   scientific   discipline,   also   automated   content   analysis   is   becoming   a   data-­‐intensive  science.   Inspiring  cases   • Usage  of  DiscoverText    in  government100   • OpinionSpace101   • Project  NOMAD102                                                                                                                             100   101   102  http://www.nomad-­‐   80  |  P a g e  
  • 81.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   Tools  on  the  market   The  market  of  opinion  mining  tools  is  crowded  with  solution  providers.  Most  of  these  applications   are   geared   towards   analysing   customers   feedback   about   products   and   services,   and   therefore   skewed   towards   sentiment   analysis   that   detects   positive/negative   feelings   by   interpreting   natural   language.   Freely  available  tools   Most   of   the   state-­‐of-­‐the-­‐art   argument   mapping   and   voter   advise   applications   are   freely   available,   because  they  derive  largely  from  academic  community  or  NGOs.  A  comprehensive  list  of  such  tools   is   available   in   and     There   are   currently   freely   available   applications   that   simply   analyse   terms   based   on   a   pre-­‐defined   glossary,  and  giver  highly  simplified  and  unreliable  results.  One  example  is          Figure  11:  Twitrratr       Another   stream   of   simple,   free   and   popular   solutions   is   the   word   visualisation.   Wordclouds   are   becoming   more   and   more   used   to   make   sense   of   large   quantities   of   information   in   a   snapshot.   Obviously,   such   tools   are   also   extremely   simplified   and   only   offer   a   visualisation   of   the   most   common  used  terms,  which  is  helpful  to  have  an  idea  of  what  the  document  is  about,  but  little  more.   Tools  such  as  provide  an  appealing  design  solution  that  can  serve  as  an  entry  level  in  the   opinion  mining  market.  They  are  therefore  important  to  involve  a  much  wider  public  in  this  kind  of   activities.   81  |  P a g e  
  • 82.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     Figure  12:  Wordclouds     Finally,  another  way  of  making  sense  of  large  amount  of  information  is  by  relying  on  human  effort,   by   crowdsourcing   and   collective   intelligence:   people   are   not   only   submitting   their   opinions,   but   actually   filtering   them   by   signalling   the   most   important   ones.   Tools   such   as   allow   customers  to  submit  feedback  and  to  rank  other  people  ideas,  thereby  allowing  the  emergence  of   the  most  popular  ideas.  These  tools  are  available  at  very  low  cost,  but  research  shows  that  they  are   effective  in  gathering  feedback  but  not  in  identifying  good  ideas,  as  voting  tends  to  focus  on  easier   and  most  popular  issues.          Figure  13:  UserVoice       82  |  P a g e  
  • 83.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   Enterprise-­‐level  software   Beside   these   simple   and   free   applications,   there   is   then   a   flourishing   market   of   enterprise-­‐level   software  for  opinion  mining  which  much  more  advanced  features.  These  tools  are  largely  in  use  by   companies   to   monitor   their   reputation   and   the   feedback   about   products   on   social   media.   In   the   government  context,  opinion  mining  has  long  been  in  use  as  an  intelligence  tool,  to  detect  hostile  or   negative   communications   (Abbasi   2007).   More   recently,   politics   has   become   a   key   area   of   applications,  as  politicians  monitor  public  opinion  on  social  media  to  understand  public  reaction  to   their  position.   Technically,   these   tools   rely   on   machine   learning   with   regard   to   identifying   and   classify   relevant   comments,   through   a   combination   of   latent   semantic   analysis,   support   vector   machines,   "bag   of   words"   and   Semantic   Orientation.   This   process   requires   significant   human   effort   aided   by   machines:   all   the   tools   on   the   market   rely   on   a   combination   of   machine   and   human   analysis,   typically   using   machines  to  augment  human  capacity  to  classify,  code  and  label  comments.   Automated  analysis  is  based  on  a  combination  of  semantic  and  statistical  analysis.  Recently,  because   of   the   sheer   increase   in   the   quantity   of   datasets   available,   statistical   analysis   is   becoming   more   important.   Key  challenges  and  gaps   Current   solutions   for   opinion   mining   and   sentiment   analysis   are   rapidly   evolving,   typically   by   reducing  the  amount  of  human  effort  needed  to  classify  comments.   Among  the  challenges  identified  we  can  select:   -­‐ -­‐ -­‐ -­‐ -­‐ -­‐ The  detection  of  spam  and  fake  reviews,  mainly  through  the  identification  of  duplicates,  the   comparison   of   qualitative   with   summary   reviews,   the   detection   of   outliers,   and   the   reputation  of  the  reviewer  (Liu  2008)   The   limits   of   collaborative   filtering,   which   tends   to   identify   most   popular   concepts   and   to   overlook  most  innovative  /  out  of  the  box  thinking   The  risk  of  a  filter  bubble  (Pariser  2011),  where  automated  content  analysis  combined  with   behavioural  analysis  leads  to  a  very  effective  but  ultimately  deviating  selection  of  relevant   opinions  and  content,  so  that  the  user  is  not  aware  of  content  which  is  somehow  different   from  his  expectations   The  asymmetry  in  availability  of  opinion  mining  software,  which  can  currently  be  afforded   only   by   organisations   and   government,   but   not   by   citizens.   In   other   words,   government   have   the   means   today   to   monitor   public   opinion   in   ways   that   are   not   available   to   the   average   citizens.   While   content   production   and   publication   has   democratized,   content   analysis  has  not.   The  integration  of  opinion  with  behaviour  and  implicit  data,  in  order  to  validate  and  provide   further  analysis  into  the  data  beyond  opinion  expressed   The   continuous   need   for   better   usability   and   user-­‐friendliness   of   the   tools,   which   are   currently  usable  mainly  by  data  analysts   Current  research   Current  research  is  focussing  on:     ● Improving  the  accuracy  of  algorithm  for  opinion  detection   ● Reduction  of  human  effort  needed  to  analyze  content   ● Semantic  analysis  through  lexicon/corpus  of  words  with  known  sentiment  for   sentiment  classification   83  |  P a g e  
  • 84.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   ● Identification  of  policy  opinionated  material  to  be  analysed   ● Computer-­‐generated  reference  corpuses  in  political/governance  field   ● Visual  mapping  of  bipolar  opinion   ● Identification  of  highly  rated  experts   Future  research:  long  term  and  short  term  issues   We  can  distinguish  between  long  and  short  term  research  efforts.  As  for  the  first  one  we  have:   • Enhanced  discoverability  of  content  through  Linked  Data   • Visual  representation   • Audio-­‐visual  opinion  mining   • Real-­‐time  opinion  mining   • Machine  learning  algorithms   • SNA  applied  to  opinion  and  expertise   • Bipolar  assessment  of  opinions   • Multilingual  reference  corpora   • Comment  and  opinion  recommendation  algorithm   • Cross-­‐platform  opinion  mining   • Collaborative  sharing  of  annotating/labelling  resources       On  the  other  hand,  for  the  long-­‐term:   ● Autonomous  machine  learning  and  artificial  intelligence   ● Usable,  peer-­‐to-­‐peer  opinion  mining  tools  for  citizens   ● Non-­‐bipolar  assessment  of  opinion   ● Automatic  irony  detection   84  |  P a g e  
  • 85.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP                           3.2.3. Visual  Analytics  for  collaborative  governance:  the  opportunities  and  the   research  challenges   Summary Overview Market  availability   Challenges  and  gaps   Current  research   Short  term  future   research   Long  term  future   research   -­‐Information   visualisation   requirements  for   business  intelligence   and  situational   awareness   -­‐Enterprise  knowledge   visualisation  linking   -­‐Online  analytical   processing  and  data   mining  -­‐Advanced  social   network  analysis  and   visualisation   -­‐Data  mining  and   interactive  visualisation   communication  of   location-­‐based   statistical  data   -­‐Information   visualisation  tools  for   high  dimensional  non-­‐ linear  data   -­‐Visual  analysis  of  data   in  spreadsheet  format   -­‐  Demographics   visualisations,  allowing   stakeholders  and   decision  makers  to  have   a  clear  picture  of  the   data  and  of  their  trends   over  time   -­‐  Legal  Arguments   visualisation:  text   analysis,  argumentation   mappings  and   visualisation  algorithms   -­‐  Discussion  Arguments   visualisation,  making   use  of  visualisation   techniques  for   visualizing  a   discussion’s  flow   -­‐Geographic   visualisation  tools   -­‐Financial  markets   monitoring  and   visualizing  in  real  time   -­‐Advanced  applications   for  security  and  defense   -­‐Close  the  loop  of   information  selection,   preparation  and   visualisation   -­‐Simultaneous  multiple   visualisation   -­‐Integration  of   visualisation  with   comments  /  wiki  /  blogs   -­‐Collaborative  platform   display   Interaction  between   visualisation  and  models   -­‐Mobile  visual  analytics   tools   -­‐Geo-­‐visualisation  of   government  data   -­‐Integration  with  opinion   mining  and  participatory   sensing   -­‐Evaluation  framework   for  visualisation   effectiveness   -­‐Visualisation   infrastructures  for  policy   modelling  issues   -­‐Re-­‐usable,   mashable  tools  for   visual  analytics   -­‐Tighter  integration   between  automatic   computation  and   interactive   visualisation   -­‐Bias  identification   and  signalling  in   visualisation   -­‐Perceptual,   cognitive  and   graphical  principles   -­‐Efficiency  of  the   visualisation   techniques  to   enable  interactive   exploration   interaction   techniques  such  as   focus  &  context   -­‐Impact  evaluation   of  visual  analytics   on  policy  choices   -­‐Learning  adaptive   algorithm  for  users   intent   -­‐Advanced  visual   analytics  interfaces   -­‐Intuitive  affordable   visual  analytics   interface  for   citizens   -­‐Development  of   novel  interaction   algorithms   incorporating   machine   recognition  of  the   actual  user  intent   and  appropriate   adaptation  of  main   display  parameters   such  as  the  level  of   detail,  data   selection,  etc.  by   which  the  data  is   presented   85  |  P a g e  
  • 86.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   Introduction  and  definition     The  explosion  in  computing  techniques  led  to  the  generation  of  a  tremendous  amount  of  data  which   are  stored  in  the  internet  and  processed  in  the  IT  infrastructures  all  over  the  world.  Some  examples   of   new   technologies   for   data   collection   are:   web   logs;   RFID;   sensor   networks;   social   networks;   social   data  (due  to  the  Social  data  revolution),  Internet  text  and  documents;  Internet  search  indexing;  call   detail   records;   astronomy,   atmospheric   science,   genomics,   biogeochemical,   biological;   military   surveillance;  medical  records;  photography  archives;  video  archives;  large-­‐scale  eCommerce.   In   managing   this   huge   amount   of   data,   when   it   comes   to   human-­‐computer   interaction   there   is   a   need  to  distil  the  most  important  information  to  be  presented  it  in  a  humanly  understandable  and   comprehensive   way.   Here   it   comes   visualisation,   which   is   a   way   to   interpret   and   translate   data   from   computer   understandable   formats   to   human   ones   by   employing   graphical   models,   charts,   graphs   and   other   images   that   are   conventional   for   humans   (Bederson   and   Shneiderman   2003).   From   one   hand   we   can   define   visualisation   as   any   technique   for   creating   create   insight,   preferably   by   allowing   users   to   interact   and   alter   with   the   visualisation   to   iteratively   solve   questions   and   form   new   questions   based   on   previous   findings.   On   the   other   hand   visualisation   can   be   defined   as   a   set   of   techniques  for  communicating  knowledge  that  can  be  supported  by  data.   In   contrast   with   visualisation   traditionally   seen   as   the   output   of   the   analytical   process,   visual   analytics 103  considers   visualisation   as   a   dynamic   tool   that   aims   at   integrating   the   outstanding   capabilities   of   humans   in   terms   of   visual   information   exploration   and   the   enormous   processing   power   of   computers   to   form   a   powerful   knowledge   discovery   environment.   In   this   view   visual   analytics  is  useful  for  tackling  the  increasing  amount  of  data  available,  and  for  using  in  the  best  way   the   information   contained   in   the   data   itself.   Moreover   visual   analytics   aims   at   present   the   data   in   way  suitable  for  informing  the  policy  making  process.   More  in  particular  the  interdisciplinary  field  of  visual  analytics  aims  at  combining  human  perception   and   computing   power   in   order   to   solve   the   information   overload   problem.   In   Thomas   and   Cooks   (2005)   definition,   visual   analytics   is   “the   science   of   analytical   reasoning   supported   by   interactive   visual   interfaces”.   Precisely   visual   analytics   is   an   iterative   process   that   involves   information   gathering,   data   preprocessing,   knowledge   representation,   interaction   and   decision   making.   The   characteristic   of   this   field   is   that   it   entails   the   association   of   data-­‐mining   and   text-­‐mining   technologies,   used   for   preprocessing   massive   amounts   of   data,   and   information   visualisation104,   which   is   useful   for   disentangling   important   from   trivial   and   useless   information.   In   a   certain   way   information   visualisation   becomes   a   tool   in   a   semi-­‐automated   analytical   process   characterized   by   the  cooperation  between  humans  and  computers,  in  which  is  the  user  who  decides  the  direction  of   the  analysis  relating  to  a  particular  task,  while  the  system  works  as  an  interaction  tool.  It  is  somehow   difficult  to  distinguish  among  information  visualisation  and  visual  analytics.  In  poor  terms  we  can  say   that   information   visualisation   handles   abstract   data   structures   such   as   trees   or   graphs,   and   finally   visual   analytics   deals   properly   with   sense-­‐making   and   reasoning.   More   in   particular   information   visualisation   is   mostly   applied   to   data   not   belonging   to   scientific   inquiry,   e.g.   graphical   representations   of   data   for   business,   government,   news   and   social   media.   Visualisation   work   does                                                                                                                             103   104  We  can  define  Information  visualisation  as  a  way  of  making  data  easier  to  understand  using  direct  sensory  experience,   rather   than   linguistic   or   logical   reasoning.   Or   in   the   words   of   Friendly,   information   visualisation   is   the   study   of   "the  visual  representation  of   large-­‐scale   collections   of   non-­‐numerical   information,   such   as   files   and   lines   of   code   in  software   systems,  library  and   bibliographic  databases,   networks   of   relations   on   the  internet,   and   so   forth".   (See   Michael   Friendly  2008)     86  |  P a g e  
  • 87.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   not  necessarily  deal  with  an  analysis  task  nor  does  it  always  use  advanced  data  analysis  algorithms.   On   the   other   hand   visual   analytics   can   be   seen   as   an   integral   approach   to   decision-­‐making,   combining  visualisation,  human  factors  and  data  analysis.  It  entails  identifying  the  best  algorithm  for   a   given   analysis   task,   to   be   integrated   with   the   best   automated   analysis   algorithms   with   appropriate   visualisation  and  interaction  techniques.   Visualisation  and  visual  analytics  should  be  considered  in  strict  integration  with  other  research  areas,   such  as  modelling  and  simulation105,  social  network  analysis,  participatory  sensing,  open  linked  data,   visual  computing.   The   disciplines   in   the   domain   of   visualisation   and   visual   analytics   include:   Human-­‐Computer   Interaction   (HCI),   Computer   Science,   Graphic   and   Information   Design,   Usability   Engineering,   Cognitive   and   Perceptual   Science,   Decision   Science,   Information   Visualisation,   Scientific   Visualisation,   Databases,   Data   Mining,   Statistics,   Knowledge   Discovery,   Data   Management   &   Knowledge   Representation,   Presentation,   Production   and   Dissemination,   Statistics,   Interaction,   Geospatial  Analytics,  Graphics  and  Rendering,  Cognition,  Perception,  and  Interaction.   As   far   the   visual   analytics   methodologies   are   concerned,   in   the   CROSSOVER   taxonomy   we   can   identify   the   following:   visualisation   of   a   single,   static,   embedded   data   set;   visualisation   of   multiple   static  data  sets;  visualisation  of  a  single  live  data  feed  or  updating  data  set;  and  finally  visualisation   of  multiple  data  points,  including  live  feeds  or  updates.   Why  it  matters  in  governance   Today’s   governments   face   the   challenge   of   understanding   an   increasingly   complex   and   interdependent   world,   and   the   fast   pace   of   change   and   increased   instability   in   all   the   areas   of   regulation  requires  rapid  decision  making  able  to  draw  on  the  wider  amount  of  available  evidence  in   real-­‐time.  How  can  visualization  and  visual  analytics  help?   • Generate   high  involvement  of  citizens  in   policy-­‐making.  One  of  the  main  applications   of   visualisation   is   in   making   sense   of   large   datasets   and   identifying   key   variables   and   causal  relationships  in  a  non-­‐technical  way.  Similarly,  it  enables  non-­‐technical  users  to   make   sense   of   data   and   interact   with   them.   For   instance,   the   GapMinder106  software   helps  to  understand  the  main  global  demographic  changes  and  raise  awareness  on  the   implications  of  sound  health  policies  in  developing  countries.     • Understand  the  impact  of  policies:  visualisation  is  instrumental  in  making  evaluation  of   policy  impact  more  effective.  For  instance,  Farmsubsidy107  helps  understanding  who  are   the  main  beneficiaries  of  the  common  agricultural  policy  by  geo-­‐referencing  the  single   beneficiary.   • Identify   problems   at   an   early   stage,   detect   the   “unknown   unknown”   and   anticipate   crisis:  visual  analytics  are  largely  used  in  the  intelligence  community  because  they  help   exploiting   the   human   capacity   to   detect   unexpected   patterns   and   connections   between   data.   Thereby   they   help   early   detection   of   potential   threats   at   an   early   stage.   For                                                                                                                             105  The  connections  between  simulation  and  visualisation  appears  even  more  clear  when  dealing  with  user  interfaces,   which  enable  the  visualisation  to  take  user  commands   106   107     87  |  P a g e  
  • 88.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   instance,   the   VisAware108  project   in   the   US   provides   situational   awareness   in   situation   of  emergencies,  helping  the  coordination  of  different  resources  involved  in  emergencies     History  and  trends   Since  from  the  beginning  of  human  history,  visualisation  has  been  an  effective  way  to  communicate   both  abstract  and  concrete  ideas.  The  appearance  of  digital  visualisation  led  to  the  development  of   graphic   hardware   as   well   as   to   a   wide   array   of   technique   used   to   visualize   data   in   a   number   of   ways   (van  Wijk,  2005).     One  of  the  best-­‐known  examples  of  visualisation  dates  back  to  the  19th  century  with  the  drawings109   by   Charles   Joseph   Minard,   who   developed   a   format   to   show   data   tied   to   a   timescale   with   a   landscape   background.   In   particular   Minard   conveyed   a   complex   series   of   events   through   various   data   measures,   explained   together   with   their   causes   and   consequences   in   a   single   graphic.   Minard's   drawings  are  applied  to  show  the  march  of  Napoleon’s  army  towards  Moscow,  starting  with  422,000   and   ending   with   10,000   men,   and   Hannibal's   crossing   of   the   Alps,   starting   with   97,000   and   ending   with  6,000  men.  The  modern  visualisation  field,  making  use  of  computer  graphics,  originated  in  the   late   1980s   with   the   studies   on   scientific   visualisation   applied   to   fluid   dynamics,   volume   visualisation,   molecular   modelling,   imaging   remote-­‐sensing   data,   and   medical   imaging   (Rosenblum   1994).   From   scientific  visualisation  took  place  some  more  recent  areas,  such  as  information  visualisation,  mobile   visualisation,   location-­‐aware   computing   and   visual   analytics.   Information   visualisation   arose   when   Robertson,   Card   and   Mackinlay   in   the   1980s   started   to   use   the   work   of   Bertin   (1967)   and   Tufte   (1983)   in   interactive   computer   applications.   Later   Shneiderman   (1996)   inter   al.   formalized   the   process   of   information   visualisation.   Finally   Ware   (2004)   emphasized   the   important   of   human   perception  in  information  visualisation.  In  parallel  with  information  visualisation  raised  the  field  of   data  mining,  aimed  at  discovering  information  hidden  in  massive  amounts  of  data.  A  characteristic  of   the  field  is  that  it  aimed  at  substituting  the  human  analysis  with  automatic  computer  operations,  not   supporting   human   perception   with   interactive   visualisation.   In   order   to   avoid   that   was   developed   the   interdisciplinary   field   of   visual   analytics,   which   combines   human   perception   abilities   with   computers’  processing  power  in  order  to  tackle  massive  amounts  of  information.  Visual  analytics  can   therefore   be   seen   as   the   combination   between   human   factors   and   data   analysis   on   one   side,   and   information   visualisation   (Keim   et   el.   2008).   Future   developments   of   visual   analytics   include   the   fields   of   enhanced   collaboration   capabilities,   more   intuitive   interaction,   support   of   non-­‐computing   devices,  as  well  as  the  integration  of  quantitative  and  qualitative  data.  In  fact  visual  analytics  require   particular  technological  advances,  as  traditional  data  mining  tools  are  unsuitable  for  some  necessary   functionalities  such  as  the  algorithm  speed  required  for  iterative  visualisation.   Inspiring  cases  in  information  visualisation  and  visual  analytics:   •  GapMinder110   •  US  Labour  Force  visualisation111   •  State  Cancer  Profiles112                                                                                                                             108   109   110   For   other   examples   please   refer   to   111   112   88  |  P a g e  
  • 89.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   •  Instant  Atlas113   • Rennes  Metropole114   • City  Dashboard115   • OECD  Better  Life  Index116     • Gain  Index117     • IBM  Many  Bills118   • Graphical  Contingency  Analysis119   • DeepCity3D120   • Vis  Sense121           Projects  in  information  visualization  and  visual  analytics:   • Jigsaw122:  visualization  for  investigative  analysis     • Ploceus123:  network-­‐based  visualization  of  tabular  data     • Dotlink360124:  visual  analytics  for  exploring  converging  business  ecosystems     • SportsVis125:  visualization  to  analyze  sports  data     • Intelligence  Analysis126:  visual  analytics  to  help  intelligence  analysts     • SellTrend127:  visualizing  temporal,  categorical  event  transactions     • Dust  &  Magnet128:  InfoVis  via  a  magnet  metaphor     • Fund  Explorer129:  stock  portfolio  diversification  through  Context  Treemaps                                                                                                                               113   114   115   116   117   89  |  P a g e  
  • 90.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • InfoCanvas130:  peripheral  information  art     • Information  Mural131:  squeezing  large  data  sets  into  small  views     • NetVizor132:  visualizing  network  topologies     • SunBurst133:  radial  space-­‐filling  views  of  hierarchies     • Tarantula134:  testing  and  debugging  large  software  systems     Policy  applications  of  visualisation  and  visual  analytics  tools   With  regard  to  the  governance  and  policy  making  context,  some  visualisation  tools  can  be  applicable   to   a   wide   array   of   issues   and   situation   (education,   environment,   public   health,   urban   growth,   national  defense,  etc.).  In  the  public  context,  visual  analytics  of  public  data  is  an  exploding  field,  with   particular   relation   to   the   open   data   movement,   in   order   to   monitor   policy   context   and   evaluate   government  policies.  Most  basic  mash-­‐up  tools  are  available  to  visualize  government.   Let  us  see  some  other  examples:   • Demographics  visualisations,  allowing  stakeholders  and  decision  makers  to  have  a  clear   picture   of   the   data   and   of   their   trends   over   time.   Visualisation   of   demographic   data   make   easier   the   design   and   evaluation   of   various   policies,   as   there   is   no   need   to   dig   through   acres   of   numbers.   In   fact   advanced   algorithms   are   able   to   create   figures   and   illustrations   easy   to   interpret.   Typical   examples   are   the   aforementioned   GapMinder   (which   embeds   visualisations   of   various   demographic   data   at   global   level),   as   well   as   Dynamic   Choropleth   Maps 135 ,   DataPlace 136 ,   Hive   Group 137 ,   Name   Voyager 138 ,   State   Cancer  Profiles139.                                                                                                                                                                                                                                                                                                                                                                                             118   120   121  http://www.vis-­‐   122   123   124   125   126   127   128   129   130   131   132   133   134   119 135   136   137     138     139     90  |  P a g e  
  • 91.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Legal  Arguments  visualisation:  text  analysis,  argumentation  mappings  and  visualisation   algorithms  can  be  applied  to  legal  documents  in  order  to  simplify  legislation  making  it   more  accessible  and  comprehensible  to  the  general  public  (Many  Bills140,  Clear  Congress   Project 141 ),  or   in   order   to  visually   represent   corroborative   evidence   (e.g.   the   tools   Carneades142,  Deflog143)   • Discussion   Arguments   visualisation,   making   use   of   visualisation   techniques   for   visualizing  the  flow  of  a  discussion  that  include  various  arguments,  in  order  to  instantly   get  awareness  of  the  topics  discussed,  as  well  as  of  the  arguments  and  the  support  such   arguments   gain.   In   this   view   visualisation   supports   all   interested   stakeholders   to   understand   the   flow   of   a   discussion,   which   is   presented   to   them   in   a   structured   and   interactive  format,  avoiding  numerous  discussion  threads.  Example  of  such  visualisation   tools   include   DebateGraph 144,   which   is   intensively   used   for   building   argumentation   maps,  as  well  as  Araucaria145,  Compendium146,  Argublogging147  and  Rationale148.   • Geovisualisation,   which   is   based   on   the   provision   of   theory,   tools   and   methods   for   visual   analysis,   synthesis,   exploration   and   representation   of   geographical   data   and   information  in  order  to  derive  problem  specific  models  and  design  task  specific  maps  for   incorporating   geographical   knowledge   into   planning   and   decision   making.   Some   examples  of  such  tools  include  ESTAT149,  GeoViz  Toolkit150,  the  geovisualisation  tools  at   the  US  National  Cancer  Institute151,  some  applications  of  InstantAtlas152.   • Advanced  visualisation  applications  used  for  security  and  national  defense.  In  this  fields,   software  advances  are  being  led  both  on  the  military  and  on  the  corporate  front.  In  fact   business   organizations   also   have   urgent   information   visualisation   requirements   that   support   their   business   intelligence   and   situational   awareness   capability,   data   mining   and   reporting   requirements.   In   this   view   many   of   the   software   innovations   are   being   targeted  at  financial  and  corporate  requirements,  but  are  also  applicable  to  the  defense   domain  due  to  common  data  mining  and  information  visualisation  challenges.  Examples   of   such   tools   are:   DataMontage153,   HoneyComb154,   Oculus   GeoTime155  and   Starlight156.                                                                                                                             140   141   143   142 144   145     146   147   148   149   150   151   152   153   154   155   156   91  |  P a g e  
  • 92.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   Other   very   interesting   examples   are   Analyst’s   Notebook Visualizer158,  adopted  by  intelligence  agencies  such  as  the  CIA.   157  is   Visual   Sentinel   • Visualisation   applications   adopted   for   financial   markets   monitoring   and   visualizing   in   real  time.  An  example  of  such  tool  is  SmartMoney159.   • Visualisation   applied   to   governmental   finances/expenditure   monitoring,   such   as   USAspending.gov160,  OffenerHaushalt161  and  Where  Does  My  Money  Go162.   Tools  on  the  market   There  is  a  massive  quantity  of  visualisation  tools  in  the  market,  both  freely  available  and  enterprise   level,  critical  for  analysts  and  researchers,  but  also  for  common  people,  is  now  available  online.   Freely  available  tools   First   of   all   we   have   visualisation   websites   useful   for   sharing   and   presenting   data,   provide   clear   context   on   important   cultural,   environmental,   social   and   economic   issue,   build   chart   and   share   visualisation  and  discoveries.  Such  examples  include  Data360163.  Moreover  there  are  “do  it  yourself”   infographic  tools  such  as  Vizify164,  Visual.ly165,  Easel.ly166  and  Vizualize.me167.     Then   we   have   data   visualisation   tools   used   for   plotting   data   on   maps,   frameworks   for   creating   charts,   graphs   and   diagrams   and   tools   to   simplify   the   handling   of   data   transforming   them   into   spreadsheets,   visual   data   mining   and   database   exploration   system,   data   visualisation   system   for   high-­‐dimensional   data,   visualisation   framework   for   animating   data.   Some   examples   of   those   tools   are:   Data   Wrangler168,   JavaScript   InfoVis   Toolkit169,   VisDB170,   Graphviz171,   IBM   OpenDX172,   Gephi173,   GeoCommons174,  Miso  Dataset175,  Polymaps176,  Tableau  Public177.                                                                                                                             157­‐product-­‐line/ibm-­‐i2-­‐analysts-­‐notebook   158­‐of-­‐the-­‐market/   159 160   161   162   163   164   165   166 167   168   169   170 171   172   173   174   175   176   177   92  |  P a g e  
  • 93.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   Tools  available  in  the  market   Apart  from  free  visualisation  tools,  there  are  also  much  more  advanced  software  which  are  used  by   firms   in   order   to   satisfy   their   information   visualisation   requirements   for   business   intelligence   support   and   situational   awareness   capability,   as   well   as   data   mining   and   reporting   requirements.   Other   uses   include   enterprise   knowledge   visualisation,   linking   knowledge   to   spatial   data,   online   analytical   processing   and   data   mining,   advanced   social   network   analysis   and   visualisation,   data   mining   and   interactive   visualisation,   communication   of   location-­‐based   statistical   data,   on-­‐line   and   batch   environment   for   business   graphics,   information   visualisation   tools   for   high   dimensional   non-­‐ linear  data,  visual  analysis  of  data  in  spreadsheet  format,  analysis  of  high  volumes  of  unstructured   text,  analysis  of  high-­‐dimensional  data  in  large  complex  data  sets  and  of  multivariate  time-­‐oriented   data.   Some   examples   of   such   software   are:   CViz   Cluster 178  visualisation,   IBM   ILOG 179  visualisation,   Spotfire180,   Survey   Visualizer181,   Infoscope182,   Sentinel   Visualizer183,   Grapheur   2.0184,   InstantAtlas185,   Miner3D186,  VisuMap187,  Drillet188,  Eaagle189,  GraphInsight190,  Gsharp191,  Tableau192.   Other  examples  of  visualisation  software  can  be  found  in   •   Key  Challenges  and  Gaps     New   tools   like   the   Many   Eyes   Word   Tree 193 ,   Treemap 194,   Tag   Cloud 195  and   Bubble   Chart 196  are   available   but   lack   interactivity.   What   is   also   missing   is   a   better   interaction   of   visualisation   approaches   and   analytical   processes   of   text   mining,   as   well   as   a   better   integration   between   new   opportunities  for  data  collection,  such  as  open  data  and  participatory  sensing,  policy  modelling  and                                                                                                                                                                                                                                                                                                                                                                                               178   179  http://www-­‐   180   181   182   183   184   185   186     187     188   189   190   191   192 193  http://www-­‐   194   195    See,   which   can   be   also   found   in   the   project   CROSSOVER   Diigo   collection   196 93  |  P a g e  
  • 94.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   visual  analytics  tools.  Most  applications  related  to  visual  analytics  of  public  data  remain  at  the  level   of  visualisation  only,  with  limited  analytical  functionalities.   Visualisation  tools  are  still  largely  design  for  analyst  and  are  not  accessible  to  non-­‐experts.  Intuitive   interfaces   and   devices   are   needed   to   interact   with   data   results   through   clear   visualisations   and   meaningful   representations.   User   acceptability   is   a   challenge   in   this   sense,   and   clear   comparisons   with   previous   systems   to   assess   its   adequacy   and   objective   rules   of   thumbs   to   facilitate   design   decisions  would  be  a  great  contribution  to  the  community.   Scalability   of   visualisation   in   face   of   big   data   availability   is   a   permanent   challenge,   since  visualisation   requires  additional  performances  with  respect  to  traditional  analytics  in  order  to  allow  for  real  time   interaction  and  reduce  latency.   Finally,  visualisation  is  largely  a  demand-­‐  and  design-­‐driven  research  area.  In  this  sense  one  of  the   main   challenge   is   to   ensure   the   multidisciplinary   collaboration   of   engineering,   statistics,   computer   science  and  graphic  design.   A   relevant   challenge   of   visualisation   and   visual   analytics   is   to   adapt   existing   techniques   to   policy   modelling:   • RelaNet   (Landesberger   et   al.   2008),   which   displays   the   network   relations   and   thereby   is   able  to  show  the  connections  and  co-­‐variances  of  the  different  opinions  overtime   • CirVis3D   (Landesberger   et   al.   2009),   which   can   visualize   clustered   opinion   snippets   as   well  as  display  time  series  in  order  to  show  the  opinion  trends  over  time   Following  Chen  (2005),  who  builds  on  Rhyne  et  al.  (2004),  we  can  enumerate  a  number  of  challenges   in  the  topic:   • Usability:   the   availability   of   low   cost,   ready   to   use   and   reconfigurable   information   visualisation   systems,   as   well   as   a   balanced   portfolio   of   general   purpose   fully   functional   information  visualisation  systems  is  used  is  crucial     • Understanding   elementary   perceptual–cognitive   tasks:   research   should   not   only   focus   on   relatively   high   level   cognitive   activities   such   as   browsing   and   searching,   or   judging   the   relevance   of   information.   Rather   it   should   primarily   focus   on   the   identification   and   de-­‐ codification   of   visualized   objects   would   be   a   fundamental   step   toward   engineering   information  visualisation  systems   • Prior   knowledge:   in   order   to   understand   the   underlying   message   in   visualized   information   users   need   a   prior   knowledge   of   how   to   operate   the   information   visualisation   system,   as   well  as  the  domain  knowledge  of  how  to  interpret  the  content   • Education  and  training:  on  the  one  hand  there  is  the  need  for  the  need  for  researchers  and   practitioners   within   the   field   of   information   visualisation   to   learn   and   share   principles   and   skills   of   visual   communication.   On   the   other   hand   potential   users   from   other   fields   must   realize  the  value  of  information  visualisation  and  how  it  might  contribute  to  their  work   • Intrinsic  quality  measures:  finding  quality  metrics  is  crucial  for  the  evaluation  and  selection   of   visual   information   advances,   and   for   understanding   to   what   extent   an   information   visualisation   design   represents   the   underlying   data   faithfully   and   efficiently,   and   preserves   intrinsic  properties  of  the  underlying  phenomenon   • Scalability:   need   for   the   adoption   of   parallel   computing   and   other   high-­‐performance   computing  techniques  in  information  visualisation   94  |  P a g e  
  • 95.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Aesthetics:   it   is   important   to   assess   how   insights   and   aesthetics   interact   in   sustaining   insightful   and   visually   appealing   information   visualisation.   What   visual   properties   make   users  think  a  graph  is  pretty  or  visually  appealing?   • Paradigm   shift   from   structures   to   dynamics:   shift   from   the   study   of   the   structure   of   visualisation   to   the   assessment   of   the   dynamic   properties   of   underlying   phenomena,   providing  built-­‐in  trend  detection  mechanisms  embedded  in  the  data  modelling  component   • Causality,   visual   inference,   and   predictions:   there   is   a   strong   necessity   for   the   elaboration   of   sensitive   and   selective   algorithms   that   can   resolve   conflicting   evidence   and   suppress   background   noises.   To   this   respect   a   great   role   is   played   by   complex   network   and   link   analysis     • Knowledge  domain  visualisation:  it  encompasses  several  of  the  aforementioned  challenges,   and   it   is   linked   to   the   fact   that   it   is   not   only   the   information   conveyed   to   be   important,   rather  is  its  structure,  which  is  a  social  construction       Current  research   • Close  the  loop  of  information  selection,  preparation  and  visualisation   • Multiple,  coordinated  views  in  visualisation/visual  analytics197   • Integration  of  visualisation  with  comments  /  wiki  /  blogs   • Collaborative  platform  display   • Interaction  between  visualisation  and  models   • Mobile  visual  analytics  tools,  e.g.  Sitegeist198   • Geo-­‐visualisation  of  government  data   • Integration  with  opinion  mining  and  participatory  sensing   • Evaluation  framework  for  visualisation  effectiveness   • Visualisation  infrastructures  for  policy  modelling  issues     A  list  of  EU  funded  projects  in  visual  analytics  include:   •  VisMaster-­‐Visual  Analytics:  Mastering  the  Information  Age199   •  VisSense-­‐   Visual   Analytic   Representation   of   Large   Datasets   for   Enhancing   Network   Security200                                                                                                                             197  See  Heer,  Jeffrey,  Fernanda  B.  Viégas,  and  Martin  Wattenberg.  2007.  Voyagers  and  Voyeurs:  Supporting  Asynchronous   Collaborative  Information  visualisation.  In  CHI  2007,  April  28–May  3,  2007,  San  Jose,  California,  USA.  See  also  the   presentation  on  social  visualisation  carried  out  by  Fernanda  Viégas  and  Martin  Wattenberg  at  the  University  of   Harvard,  April  13  2009   198­‐uncover-­‐the-­‐data-­‐around-­‐you/   199  http://www.visual-­‐   200   95  |  P a g e  
  • 96.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   •  CUBIST-­‐  Combining  and  Uniting  Business  Intelligence  and  Semantic  Technologies201   •  WATTALIST  –  Modelling  and  Analysing  Demand  Response  Systems202   •  CODE  –  Commercially  empowered  Linked  Open  Data  Ecosystems  in  Research203   •  SemSeg-­‐4D  Space-­‐Time  Topology  for  Semantic  Flow  Segmentation204       Future  research:  long  term  and  short  term  issues   Short-­‐term  research   • Reusability   of   mashup   tools   (mashup   is   a   web   application   which   combines   data   from   one  or  more  sources  into  a  single  integrated  tool  or  application)  for  visual  analytics   • Tighter  integration  between  automatic  computation  and  interactive  visualisation,  which   consists   in   the   availability   of   complex   and   powerful   algorithms   that   allow   for   manipulating   the   data   under   analysis,   transforming   it   in   order   to   feed   suitable   visualisations   • Bias  identification  and  signalling  in  visualisation   • Techniques  and  algorithms  for  creating  effective  visualisation  tools  based  on  perceptual   psychology   (dealing   with   the   process   by   which   the   physical   energy   received   by   sense   organs   forms   the   basis   of   perceptual   experience),   cognitive   science   (focusing   on   how   information  is  represented,  processed,  and  transformed)  and  graphical  principles   • Visualisations   enabling   interactive   exploration   techniques   such   as   focus   &   context,   in   order  for  the  viewers  to  be  able  to  see  the  object  of  primary  interest  presented  in  full   detail   while   at   the   same   time   getting   a   overview–impression   of   all   the   surrounding   information  —  or  context  —  available   • Exploiting  visualisation  as  a  medium  to  engage  citizens  in  policy-­‐related  complex  matter   • Visualisation   as   a   way   to   provide   (persuasive)   feedback   and   change   in   attitudes,   opinions,  behaviors   • Visualisation  as  a  medium  for  grassroots/crowd-­‐sourced  participation,  collaboration  on   data-­‐related  issues     • Impact  evaluation  of  visual  analytics  on  policy  choices   • Research  in  making  visualisation  accessible  for  non-­‐experts       Long-­‐term  research                                                                                                                             201   202   203   204     96  |  P a g e  
  • 97.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Learning  adaptive  algorithm  for  users  intent.  Note  that  learning/adaptive  algorithms  are   defined   as   being   capable   to   automatically   change   behaviour   based   on   its   execution   context   (data   handled   by   the   algorithm,   configuration   parameters   of   the   runtime   environment,  resources  used)  in  order  to  obtain  optimal  performances   • Advanced  visual  analytics  interfaces:  visual  interfaces  in  which  neither  the  analytics  nor   the   visualisation   needs   to   be   advanced   in   itself   but   synergy   between   automation   and   visualisation  is  in  fact  advanced   • Intuitive  and  affordable  visual  analytics  interface  for  citizens   • Development  of  novel  interaction  algorithms  incorporating  machine  recognition  of  the   actual  user  intent  or  of  the  actual  relevance  for  the  user  and  appropriate  adaptation  of   main   display   parameters   such   as   the   level   of   detail,   data   selection,   etc.   by   which   the   data  is  presented   97  |  P a g e  
  • 98.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   3.2.4. Serious  Gaming  for  Behavioural  Change   Introduction  and  definition   So   far,   collaborative   ICTs   have   dramatically   augmented   the   capacity   of   people   to   connect   and   collaborate.   Yet,   less   impact   has   been   achieved   in   terms   of   actual   change   and   action,   as   most   collaboration   remain   confined   to   an   elite   of   highly-­‐motivated   individuals   and   faces   the   traditional   limits   of   human   attention   and   motivation.   As   illustrated   in   other   challenges,   ICT   can   improve   data   collection   and   analysis,   but   if   attention   and   motivation   are   not   present,   little   impact   can   be   achieved.  This  challenge  depicts  ICT  solutions  that  enable  behavioural  change  and  action.  Even  when   citizens  and  government  are  fully  aware  of  necessary  policy  choices,  they  might  irrationally  choose   short-­‐term  benefits.     Simulation  and  serious  gaming  (also  known  as  interactive  learning  environments)  offer  opportunities   to  impact  on  personal  incentives  to  action  and  showing  long-­‐term  and  systemic  effects  of  individual   choices,   thereby   lowering   the   engagement   barrier   to   collaborative   governance   and   augmenting   its   impact.   In   particular,   serious   games   have   been   developed   for   educational   purposes   and   raising   awareness  on  particular  issues  while  not  requiring  high  levels  of  engagement.   Simulation   tools   enable   users   to   see   the   systemic   and   long-­‐term   impact   of   their   action   in   a   very   concrete   and   tangible   form,   thereby   encouraging   more   responsible   behaviour   and   long-­‐term   thinking.   Gaming   engages   users   through   the   “fun”   and   “social”   dimension,   thereby   providing   incentives  towards  action.  Feedback  and  simulation  systems  include  both  individual  and  government   behaviour,  thereby  allowing  policy-­‐makers  and  citizens  to  detect  the  impact  of  both  individual  and   policy  choices.   Engagement   of   domain   experts   is   a   crucial   issue   for   building   reliable   games   and   simulation   tools.   Toolkits  and  modules  enable  a  wider  audience  of  stakeholders  to  take  a  direct,  active  role  in  games   development,   thereby   enabling   all   relevant   knowledge   to   be   elicited   and   captured   by   the   simulation   and  gaming  scenarios  and  models.  Pre-­‐built   toolkit   enables   the   creation   directly   by   thematic   experts   and  not  by  technology  experts.     Why  it  matters  in  governance   Most   applications   of   simulation   and   gaming   are   developed   into   the   context   of   education   and   learning,  while  more  interactive  feedback  producing  systems  have  been  applied  to  personal  health   and   energy   conservation.   The   specific   challenges   of   gaming   for   public   policy   awareness   and   action   are   currently   less   researched,   but   are   very   specific   because   of   their   large-­‐scale   interaction   and   systemic  effects  of  individual  behaviour,  which  characterised  this  field.     Furthermore,  the  availability  of  a  simulation  toolkit  is  necessary  to   empower  a  diverse  and  inclusive   simulation  landscape,  where  the  most  diverse  set  of  ideas  can  be  influential  and  listened  to.       Recent  trends   Simulation   and   gaming   have   started   to   be   applied   in   different   policy   contexts   in   order   to   engage   wider   audiences.   Games   are   developed   “on   purpose”,   by   highly   skilled   developers,   in   the   public   sector   and   by   civil   society,   therefore   requiring   significant   investment   and   without   the   specific   thematic   knowledge   of   the   field.   Furthermore,   existing   serious   games   lack   flexibility   to   allow   for   unpredictable  developments  and  non-­‐linear  behaviours,  where  scenarios  evolve  and  adapt  to  users   choices   rather   than   being   rigidly   prescribed.   Commercial   solutions   that   turn   long-­‐term   effects   into   short-­‐term  feedback  are  available,  but  still  lack  usability  as  well  as  the  fun  dimension  of  games  and   98  |  P a g e  
  • 99.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   finally   require   high   levels   of   engagement.   They   are   designed   for   individual   feedback   and   do   not   cover  the  complexity  of  systemic  interactions,  which  are  typical  of  public  governance  issues.     To   sum   up,   serious   gaming   is   still   requiring   high   level   of   engagement,   and   progress   is   needed   in   terms   of   usability   and   appeal   in   order   to   reach   “casual   gamers”   including,   immersive   and   emotion   aware  games.     Current  practice   •   Purpose-­‐built   gaming   and   simulation   for   understanding   of   policy   issues   and   of   individual   behaviour     Public  Policy  Applications   Simulation   and   gaming   can   be   useful   to   policy   makers   in   the   following   terms   (Mayer   et   al.   2004   ,   Bots  and  van  Daalen  2007  ):   •   Research  and  analyse  a  policy  issue  when  it  is  not  feasible  to  tackle  the  real  system  (due  to   time   constraint   or   just   because   it   does   not   exists)   or   to   include   human   behaviour   by   way   of   a   computer  model  (due  to  unrealistic  assumptions  such  as  perfect  rationality).  In  this  view  the  game   becomes  a  laboratory  which  can  produce  a  great  deal  of  data  which  provide  useful  insights   •   Design  alternative  solutions  to  a   problem  analyse  and  assess  the  possible  consequences  of   the  alternative  solutions  in  order  to  recommend  a  course  of  action  for  the  policy-­‐maker.  In  this  view   the   game   can   be   seen   as   a   virtual   design   studio   useful   to   boost   out-­‐of-­‐the-­‐box   thinking   about   alternative  solutions  to  a  policy  issue,  and  also  to  ponder  recommendations’  consequences     •   A  game  can  be  used  to  provide  strategic  advice  acting  as  a  virtual  practice  ring  in  which  the   policy  maker  can  rehearsal  different  strategies.    A  typical  example  of  such  kind  is  given  by  the  war   games,  in  which  the  other  players  act  as  sparring  partners  for  the  policy  makers,  playing  the  role  of   another  stakeholder  as  opportunistically  as  possible   •   Many   policy   issues   require   mediation   so   that   it   is   necessary   to   seek   for   consensus   among   stakeholders.  This  can  be  done  by  putting  the  players  around  a  virtual  negotiation  table  by  the  mean   of   a   mediation   game.   In   this   way   the   changes   in   attitude   and   the   discovery   of   new   opportunities   for   conflict  resolution  are  eased  by  the  interaction  among  stakeholders  during  the  game   •   Normally   experts   and   elites   are   involved   in   the   policy-­‐making   process,   while   citizens   and   ordinary   people   are   completely   neglected.   However,   by   defining   virtual   consultation   forums   it   is   possible   to   allow   equal   access   for   all   the   actors   carrying   views   and   opinions,   which   would   have   been   otherwise   disregarded.   In   this   respect   using   games   and   simulations   bears   and   advantage   given   by   the  fact  that  ordinary  people  can  focus  and  express  themselves  more  easily  when  playing  a  role   •   Clearly  ethical  questions  and  opinions  have  a  great  influence  on  the  policy  making  process.   Games   and   simulations   can   be   used   to   clarify   the   values   and   arguments   behind   a   point   of   view.   While   in   ordinary   political   debate   values   remain   implicit,   by   creating   a   virtual   parliament   it   is   possible   to   make   them   explicit.   Furthermore   gaming   and   simulations   can   be   used   to   magnify   positions   and   opinions   of   stakeholders,   so   that   the   game   can   be   designed   to   reward   players   for   quality  and  clarity  of  argumentation   Moreover   readapting   the   taxonomy   of   Sawyer   and   Smith   simulation   and   serious   gaming   can   be   useful  in  the  following  domains  (cross-­‐referenced  with  game  objectives):   •   Public   sector   and   NGOs:   public   health   education   and   mass   casualty   response   (games   for   health);   political   games   (advergames);   employee   training   (games   for   training);   provide   info   to   the   99  |  P a g e  
  • 100.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   public   (games   for   education);   data   collection/planning     (games   for   research);   strategic   and   policy   planning,  spatial  planning  (games  for  producing);  diplomacy  and  opinion  research  (games  as  work)   •   Defence:   rehabilitation   and   wellness   (games   for   health);   recruitment   and   propaganda   (advergames);   soldier/support   training   (games   for   training);   school   house   education   (games   for   education);   wargames   and   planning   (games   for   research);   war   planning   and   weapons   research   (games  for  producing);  command  and  control  (games  as  work)   •   Healthcare:   cyber   therapy/exergaming   (games   for   health);   public   health   (advergames);   policy  and  social  awareness  campaigns  (games  for  training);  training  games  for  health  professionals   (games  for  education);  games  for  patient  education  and  disease  management  (games  for  research);   visualization   and   epidemiology;   biotech   manufacturing   and   design   (games   for   producing);   public   health  response  planning  and  logistics  (games  as  work)   •   Education:  inform  about  diseases/risks  (games  for  health);  social  issue  games  (advergames);   train   teachers/workforce   skills   (games   for   training);   learning   (games   for   education);   computer   science   and   recruitment     (games   for   research);   P2P   Learning   (games   for   producing);   distance   learning    (games  as  work)       Inspiring  cases     Let  us  present  now  some  inspiring  cases  of  serious  games  applied  to  policy  making:   •   SimHealth:   The   National   Health   Care   Simulation   is   a   management   simulation   of   the   U.S.   Healthcare  system  released  during  Congressional  debates  on  the  Clinton  health  care  plan   •   SimCity   2013:   is   an   upcoming   city-­‐building/urban   planning   simulation   computer   game   allowing   allows   players   to   visualize   data,   such   as   pollution   and   water   distribution,   which   will   be   realised  in  February  2013   •   City  One:  the  game  teaches  industry  professionals  and  civil  servants  the  real-­‐world  planning   in  fields  such  as  optimization  of  banking,  retail,  energy  and  water  solutions   •   Democracy   2:   government   simulation   game   in   which   the   player   acts   as   the   president   or   prime   minister   of   a   democratic   government   introducing   and   altering   policies   in   areas   such   as   tax,   economy,  welfare,  foreign  policy,  transport,  law  and  order  and  public  services   •   Close  Combat  Marines:  serious  game  for  military  training  purposes,  with  particular  reference   to  the  United  States  Marine  Corp     •   Incident  Commander™  NIMS-­‐compliant  training  tool  for  Homeland  Security:  in  this  game  the   player   mimics   the   role   of   incident   commander   in   case   of   natural   or   manmade   disaster,   terrorist   attack  or  hostage  situation.  Application:  US  Department  of  Justice  officers’  training   •   Virtual   Battlespace   Systems   2:   this   is   an   interactive   military   simulator   developed   for   the   United   States   Marine   Corp   and   the   Australian   Defence   Force   to   meet   the   individual   needs   of   military,   law   enforcement,   homeland   defence,   loadmaster,   and   first   responder   training   environments   •   Pulse!!    Virtual  Clinical  Learning  Lab  for  Health  Care  Training:  the  game  recreates  a  lifelike,   interactive,   virtual   environment   in   which   civilian   and   military   heath   care   professionals   practice   clinical  skills  in  case  of  catastrophe  or  terrorist  attack   100  |  P a g e  
  • 101.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   •   Levee  Patroller:  immersive  3D  game-­‐based  environment  to  train  levee  inspection  knowledge   and  skills,  in  order  to  be  prepared  to  cope  with  unexpected  flooding   •   game-­‐based   learning   environment   allowing   players   to   experience   and   debrief   some  of  the  complexities  involved  in  large-­‐scale  urban  projects.  Application:  development  plan  for   Scheveningen-­‐Harbour  of  The  Hague   •   Simport-­‐Maasvlakte   2:   computer-­‐supported   multi-­‐player   simulation   game   that   mimics   the   real  processes  involved  in  planning,  equipping  and  exploiting  the  new  area  in  the  Port  of  Rotterdam   •   Pro  Rail:  capacity  optimization  of  a  complex  infrastructural  network,  in  this  case  the  Dutch   railways.  Applications:  cargo  capacity  management,  opening  of  the  Vecht-­‐bridge,  increase  traffic  on   the  A2-­‐corridor   • • • • Win   Manager:   online   multiplayer   negotiation   business   game   in   which   players   conduct   a   sequence   of   bilateral   negotiations   pursued   through   private   threads   on   the   general   game   board   Management   Business   Game:   business   game   focussed   on   the   simulation   of   a   company’s   management  in  a  competitive  market,  which  can  be  played  both  online  and  offline   Management   Utilities   Euroshop:   management   of   a   chain   of   retail   stores   selling   electronic   products,   through   which   players   identify   the   relationships   between   management   issues   and   competitive  market  factors   Shadow  Government:  serious  game  based  on  the  gamification  of  real  countries,  systems,  and   worldwide   events.   Based   on   System   Dynamics,   customized   at   the   country   level,   it   allows   players  to  test  several  policy  interventions  and  evaluate  their  impacts   101  |  P a g e  
  • 102.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     Key  challenges  and  gaps   Following  Mayer  (2009)  we  can  identify  the  following  challenges  and  gaps:   •   Cultural  changes  concerning  the  interaction  between  science  in  politics,  democracy.  Changes   in   the   role   of   elites,   activism   and   citizens’   participation,   as   well   as   the   recent   emergence   of   game   cultures   •   Changes  in  public  policy  making  perception,  i.e.  from  rational  comprehensive  to  political  and   incremental     •   How  natural  and  human-­‐caused  events  can  influence  the  political  agenda  (climate  warming,   pollution,  depletion  of  natural  resources,  terrorism)   •   Institutional  changes  and  the  emergence  of  new  industrial  or  institutional  actors   •   Technological   innovation   in   computing   and   simulation   modelling,   such   as   agent-­‐based   models,  cellular  automata,  or  virtual  game  worlds.       Moreover  following  IDATE  we  can  display  the  following  key  challenges  regarding  in  particular  serious   gaming:   •   Restructure  the  game  in  order  to  cope  with  specific  purposes  and  broaden  the  audience   •   Innovate  the  existing  business  models   •   Automating   a   portion   of   the   production   process,   such   as   for   example   the   integration   of   sector-­‐specific  elements   •   Try   to   persuade   reluctant   users   and   create   sector-­‐targeting   serious   gamins   and   persuading   reluctant  users   •   Investing  in  all  connected  platforms   Current  research   •   Kit-­‐based  serious  games   •   Integration  between  policy  models  and  simulation   •   Design   of   appealing,   adaptive   and   context-­‐aware   interfaces.   Impact   of   simulation   and   gaming  on  individual  behaviour   •   Unconscious  impact  of  feedback  systems     Research   disciplines:   human-­‐computer   interaction,   sensors,   information   visualisation,   sensor   design,   psychology,  pedagogy,  public  policy     Future  research:  long  term  and  short  term  issues   Short-­‐term  research   •   Citizens-­‐  and  experts-­‐generated  gaming   •   Immersive  interfaces   102  |  P a g e  
  • 103.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   •   Large-­‐scale  collaboration  in  development   •   Casual  serious  gaming   •   Ethical  issues  in  serious  gaming   •   User-­‐controlled  simulation  and  gaming   •   Non-­‐linear  and  adaptive  scenarios  for  gaming  in  policy  context   •   Integrated  analysis  of  information  and  behavioural  change   •   Impact  of  simulation  and  gaming  on  systemic  behaviour     Long-­‐term  research   •   Augmented  reality  citizens-­‐generated  gaming  and  simulation   •   Ubiquitous  feedback  systems  on  public  governance   •   Model  and  display  long-­‐term  systemic  effects  of  individual  choices  on  public  policy  topics   •   Interplay  between  different  feedback  systems  and  other  information  outputs           3.2.5. Linked  Open  Government  Data   The   notion   of   Government   Data   concerns   all   the   information   that   governmental   bodies   produce,   collect   or   pay   for.   This   could   include   geographical   data,   statistics,   meteorological   data,   data   from   publicly   funded   research   projects,   and   digitized   books   from   libraries.   In   this   respect   the   definition   of   Open   Public   Data   is   applicable   when   that   data   can   be   readily   and   easily   consulted   and   re-­‐used   by   anyone   with   access   to   a   computer.   In   the   European   Commission's   view   'readily   accessible'   means   much   more   than   the   mere   absence   of   a   restriction   of   access   to   the   public.   Data   openness   has   resulted   in   some   application   in   the   commercial   field,   but   by   far   the   most   relevant   applications   are   created   in   the   context   of   government   data   repositories.   With   regard   to   linked   data   in   particular,   most   research   is   being   undertaken   in   other   application   domains   such   as   medicine.   Government   starts  to  play  a  leading  role  towards  a  web  of  data.  However,  current  research  in  the  field  of  open   linked  data  for  government  is  limited.   Following   the   Open   Government   Working   Group   Meeting   in   Sebastopol 205  and   the   Sunlight   Foundation206,  there  is  a  set  of  principles  according  to  which  data  can  be  considered  open:   • • • • • Data  must  be  completed,  i.e.  no  part  of  them  should  be  omitted  due  to  security,  privacy   or  privilege  limitations   Data  must  be  primary,  disaggregated  and  not  modified,  and  must  be  published  with  the   finest  possible  level  of  granularity   Data  must  be  timely  as  their  value  is  time-­‐relevant   Data  must  be  accessible  to  the  widest  range  of  users  and  purposes   Data  formats  must  not  under  exclusive  control  of  an  entity                                                                                                                             205­‐open-­‐data-­‐principles   206 103  |  P a g e  
  • 104.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • • • • • Data  should  not  be  subject  to  any  copyright   Data  must  be  machine-­‐processable   Data  access  must  be  non-­‐discriminatory   Data  must  be  permanent,  so  that  their  embedded  information  is  available  over  time   Data  must  be  cheaply  accessible   In  the  same  way  according  to  Davies  et  al.207  engaging  open  data  should:   • • • • • Be  demand  driven   Put  data  in  context   Support  conversation  around  data   Build  capacity,  skills  and  networks   Collaborate  on  data  as  a  common  resource   Moreover   according   to   Vander   Sande   et   al.  208,   publishing   data   leads   to   more   transparency,   new   businesses,   better   evidence-­‐based   policy   making   and   increased   public   sector   efficiency   only   if   the   different  actors  in  the  chain  have  co-­‐ownership  of  the  data  and  be  able  to  participate  directly  in  its   correction.  In  this  sense  free  licensing  and  shared  platform  to  publish  and  offer  feedback/corrections   directly  to  the  data  are  crucial.   Linked  open  government  data  are  valuable  for  a  number  of  reasons.  Firstly,  openness  in  government   data   is   important   for   the   economic   reasons.   For   instance,   the   Open   Data   Strategy   for   Europe   launched   by   the   European   Commission   is   expected   to   deliver   a   €40   billion   boost   to   the   EU's   economy  each  year.  More  in  general,  open  government  data  are  important  for  participatory  decision   making:   •                  Promotion  of  transparency  concerning  the  destination  and  use  of  public  expenditure   •                  Improvement  in  the  quality  of  policy  making,  which  becomes  more  evidence  based   •                  Display  the  full  economic  and  social  impact  of  information,  and  create  services  based   on  government  data   •                  Increase  in  the  collaboration  across  government  bodies,  as  well  as  between   government  and  citizens   •                  Permits  new  added-­‐values  services  to  come  into  existence   •                  Increase  the  awareness  of  citizens  on  specific  issues,  as  well  as  their  information  about   government  policies   •                  Promote  accountability  of  public  officials   • Very   important   examples   are   given   within   the   scope   of   the   Open   Government   Initiative 209  carried   out   by   the   Obama   Administration   for   promoting   government   transparency   on   a   global   scale:Data.gov210:   platform   which   increases   the   ability   of   the   public   to   easily   find,   download,   and   use   datasets   that   are   generated   and   held   by   the   Federal   Government.   In   the   scope   of   US   and   India   have   developed   an   open   source   version   called   the   Open   Government   Platform 211  (OGPL),   which   can   be   downloaded   and   evaluated   by   any   national   Government   or   state   or   local   entity   as   a   path  toward  making  their  data  open  and  transparent   • USAspending.gov212:   it   is   a   searchable   website   displaying   for   each   Federal   award   the   name  of  the  entity  receiving  the  award,  the  amount  of  the  award,  information  on  the   award,  and  the  location  of  the  entity  receiving  the  award                                                                                                                             207   209   210 211   212   208 104  |  P a g e  
  • 105.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • • • 213 :   HTML   edition   of   the   Federal   Register   to   make   it   easier   for   citizens  and  communities  to  understand  and  get  informed  about  the  regulatory  process 214 :   website   providing   a   window   of   US   Government   Administration   effort  to  improve  performance  and  accountability,  in  order  to  create  a  government  that   is  more  effective,  efficient,  innovative,  and  responsive   IT   Dashboard215:   website   enabling   federal   agencies,   industry,   the   general   public   and   other  stakeholders  to  view  details  of  federal  information  technology  investments     At   the   European   level   we   have   the   repository   of   applications   making   use   of   open   data:   publicdata.eu216.  At  the  European  national  level  the  initiatives  include:     • • • • • United  Kingdom:,  which  collects  data  from  5,400  datasets  available,  from   all   central   government   departments   and   a   number   of   other   public   sector   bodies   and   local  authorities.     Italy:,   which   is   an   open   data   portal   allowing   citizens,   developers,   firms   and   public  administrations  to  make  use  of  the  public  administration  information  stock   Spain:  datos.gob.es219,  the  national  portal  for  managing  and  organizing  the  Catalogue  of   Public  Information  of  the  General  State  Administration   Ireland: 220 ,   providing   standard   documentation   on   recurring   official   statistics  and  links  to  where  they  can  be  found   Netherlands: 221 ,   the   central   access   point   to   all   information   about   government  organizations  of  the  Netherlands                                                                                                                             213   215   216   217   218­‐il-­‐portale-­‐dei-­‐dati-­‐aperti-­‐della-­‐pa   219   220   221 214 105  |  P a g e  
  • 106.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   At  transnational  level  there  are  the  World  Bank222,  United  Nations223,  REEP224  and  Open  Knowledge   Foundation225  portals.   As  highlighted  by  the  experience  of  the  Open  Corporates226,  which  turn  the  freely  available  raw  data   into  something  genuinely  useful  that  customers  will  be  prepared  to  pay  for,  open  data  are  valuable   also  for  the  private  sector.     Figure  14    Open  Data  Business  Model  (source:  Istituto  Superiore  Mario  Boella)       In  this  case  the  data  can  generate  revenue  in  a  number  of  ways:   •                Subscriptions  or  royalties;   •                The  so-­‐called  “freemium”  model  where  a  basic  service  of  offered  for  free  but  with   charges  for  premium  services;   •                  Advertising  by  third  parties;   •                Cross  subsidy;   •                By  offering  services  that  are  cheaper  and  more  efficient  to  outsource.                                                                                                                             222   224   225   226   223 106  |  P a g e  
  • 107.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP     Looking  at  the  best  practices  and  examples  of  Linked  Open  Government  Data,  it  is  possible  to  refer   to  diagram  maintained  by  Richard  Cyganiak  (DERI,  NUI  Galway)  and  Anja  Jentzsch  (HPI),  which  is  a   visualization  of  the  key  LOD  providers  and  their  linkages227:        Figure  15  -­‐LOD  providers  and  their  linkages           Three  other  inspiring  cases  are:   • The   clean   energy   information   gateway 228 ,   which   makes   use   of   LOD   for   providing   comprehensive   clean   energy   country   profiles   so   that   users   can   access   the   highest   quality   information   in   a   visually   appealing   fashion.   By   using   small   organizations   can   share   responsibilities,   as   they   are   not   required   to   maintain   large   databases.   Moreover   the   information   is   directly   linked   to   data   providers,   so   that   updates  take  place  immediately.   • Open   Energy   Information   (OpenEl) 229 ,   a   collaborative   knowledge-­‐sharing   platform   providing  open  and  free  access  to  energy  related  models,  tools  and  data.  The  business   benefits  of  using  this  system  stem  from  the  fact  that  a  small  organization  not  having  a   huge  team  of  people  for  maintaining  a  large  database  with  information  of  clean  energy,   can   obtain   the   same   amount   of   knowledge   making   use   of   an   overview   of   a   variety   of   energy-­‐related  and  country-­‐specific  topics.  Moreover,  given  the  direct  link  to  the  data   providers’  information,  any  update  occurs  in  real  time.   • The   UK   official   government   archive,   which   offers   access   to   all   published   UK   legislation   so   that   it   can   be   shared  by   citizens   and   businesses.   The   dataset                                                                                                                             227  http://lod-­‐   229  ttp://   228 107  |  P a g e  
  • 108.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   covers   more   than   800   years   and   includes   the   most   recent   changes   in   legislation.   merges   the   contents   of   the   Office   of   Public   Sector   Information   website   and   the   Statute   Law   Database   in   ordered   to   provide   UK   public   and   local   acts,   church  instruments,  ministerial  orders  and  acts  of  the  parliament.   • • • • • • • The  tool  publicspending.gr230,  which  takes  data  from  a  variety  of  sources  and  from  the   combined  data  the  system  is  able  to  derive  graphs  showing  where  public  money  is  being   spent   and   which   departments   are   spending   it.   This   is   the   first   linked   open   data   application  in  Greece,  where  it  can  used  to  aid  to  policy  making  and  transparency   Another   interesting   case   is   “Where   Does   My   Money   Go?”231,   which   shows   how   daily   taxes  are  allocated  among  the  different  functions  of  the  government.   UK   Crime   Map 232 ,   which   has   prompted   a   change   in   the   way   police   resources   are   prioritized,  and  is  widely  used  by  the  government  itself  which  is  benefitting  from  much   more  efficient  access  to  information   The  BudgIT233  platform,  which  turned  the  Nigerian  budget  into  an  interactive  document,   complete  with  commentary  channels  via  the  Web  and  SMS   The   DERI's   Galway   Volvo   Ocean   Race 234  app   for   Android   and   iPhone,   created   by   converting   various   data   sets   into   linked   data   and   then   enrich   that   data   through   crowdsourcing.  DERI  was  used  to  classify  350  apps,  showing  that  the  majority  of  apps:   • Have  been  produced  by  individuals  rather  than  commercial  companies     • Are  Web  based   • Combine  OGD  with  maps   • Rely  on  static  data  sets  rather  than  real  time  data   • Use  a  single  data  set,  rather  than  mixed  data   The   Open   Culture   Data   project   (Open   Cultuur   Data) 235 ,   presenting   the   results   of   a   hackathon.   The   winning   entry   made   use   of   a   video   dataset   and   smartphone   capabilities   to  match  a  person's  location  with  video  taken  in  a  given  area   GLAMs236  (galleries,  libraries,  archives,  museums),  which  provides  open  access  to  the   cultural  heritage         Some   other   interesting   tools   are   the   RDF   Data   Cube   Vocabulary 237 ,   the   Data   Cube   faceted   browser238,  Openpolis239,  Nosdeputes240,  Virtueel  Platform241       Current  challenges:  open  data  and  opinion  mining   There  is  lot  of  effort  for  extracting  public  opinion  and  sentiment  towards  policies242.  The  problem  is                                                                                                                             230   232   233   234   235   236   237­‐data-­‐cube/   238   239   240   241­‐for-­‐amsterdam-­‐zoekt-­‐nieuwe-­‐open-­‐data-­‐apps/   242   231 108  |  P a g e  
  • 109.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   that   the   tools   that   are   better   at   extracting   real   data   from   social   media   content   are   of   course   expensive.  In  this  respect  the  future  challenges  for  opinion  mining  applied  to  social  media  are:   •                How  to  reduce  human  efforts;   •                Identification  of  good  ideas;   •                Finding  necessary  investments;   •                How  to  improve  usability  of  tools.   We  also  have  to  notice  that  most  of  social  media  interaction  is  not  carried  out  in  public:  for  example   in  Facebook  only  open  discussion  pages  can  provide  information  for  sentiment  analysis  and  opinion   mining.     Current  research  topics   Integration  of  open  government  data  (OGD)  and  social  media  data  (SMD):  policy  makers   will   soon   be   able   to   see   the   subjective   reaction   to   objective   changes   through   a   dashboard  that  is  powered  by  linked  data   •             3.2.6. Collaborative  Governance       Introduction  and  definition     While   all   challenges   provide   opportunities   for   a   more   effective   large-­‐scale   collaboration   in   public   action,   the   relevant   institutional   design   is   far   from   being   introduced.   The   formal   inclusion   of   citizens   input  in  the  policy-­‐making  process,  the  deriving  institutional  rules,  the  legitimacy  and  accountability   framework   are   all   issues   that   have   so   far   been   little   explored.   Instant,   open   governance   implies   a   substantial   increase   in   feedback   loops   that   are   of   a   different   scale   with   respect   to   the   present   context.  Any  system  stability  is  affected  by  the  number,  speed  and  intensity  of  feedback  loops,  and   the  institutional  context  has  been  designed  for  less  and  slower  loops.     The   definition   and   design   of   public   sector   role   is   being   directly   affected   by   the   radical   increase   in   bottom-­‐up   collaboration,   deriving   from   the   lower   cost   of   self-­‐organisation.   There   are   also   important   questions   to   be   answered   –   where   does   the   legitimacy   come   from,   how   to   gain   and   maintain   the   trust  of  users,  how  to  identify  the  users  online.  There  is  also  a  very  important  issue  of  how  to  take   into   the   account   the   diversity   of   the   standpoints,   i.e.   how   to   achieve   a   consensual   answer   to   controversial   social   issues,   especially   when   we   do   not   offer   alternatives   (ready-­‐made   options)   but   start   from   an   open   question   and   work   throughout   different   options   proposed   by   participants.   Furthermore,   the   trade-­‐off   between   direct   or   representative   model   of   democracy   will   have   to   be   analysed   in   this   context.   It   is   far   from   being   proved   that   the   open   and   collaborative   governance   is   really   inclusive   and   representative   of   all   the   social   groups,   including   the   disadvantaged   and   of   all   standpoints.  There  is  a  visible  risk  that  online  collaboration  increases  the  divide,  rather  than  reduces   it.  However,  in  the  current  situation,  the  circles  where  the  policy  proposals  are  designed,  amended   and  ranked  hierarchically  are  very  small  and  composed  by  leaders  of  political  parties,  top-­‐level  civil   109  |  P a g e  
  • 110.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   servants,   CEOs   of   large   firms.   On-­‐line   collaborative   tools   would   broaden   these   circles   to   all   those   that  have  a  competence  in  the  field  being  discussed,  and  have  an  ability  to  elaborate  an  argument.   The   management   of   institutional   bodies   is   changing:   innovative   ideas   and   insight   coming   from   employees   and   citizens   are   key   resources   to   be   exploited,   and   meritocracy   and   transparency   are   entering  a  once  stable  and  conservative  workforce.  Enhanced  collaboration  with  citizens  and  private   third   parties   should   be   accompanied   by   adequate   legal   and   accountability   frameworks,   mapping   incentives  to  participation  and  enabling  business  models  for  different  stakeholders.     The  privacy  paradigm  is  changing  and  appropriate,  more  dynamic  frameworks  have  to  be  designed,   taking   into   account   the   willingness   of   citizens   to   share   information   and   at   the   same   time   ensuring   their  full  awareness  of  the  implications  and  their  control  over  the  data  usage.       Recent  trends   The   current   status   is   characterized   by   practice-­‐driven   implementation,   accompanied   by   little   scientific  reflection.  Guidelines  and  soft  regulation  are  being  created  from  scratch  and  by  building  on   other   institutions   examples.   The   development   of   collaborative   governance   is   growing   rapidly   without  an  appropriate  reference  framework.       Public  Policy  Application       As  it  is  widely  recognized,  policy  issues  of  our  age  can  be  addressed  only  through  the  collaboration   of   all   the   components   of   the   society,   including   the   private   sector   and   individual   citizens.   In   this   view   the  advantages  in  collaborative  governance  are  given  by:   • Effectiveness  and  efficiency  in  the  delivery  of  programs   • Professional  development  /  capacity  building   • Better  needs  assessment  and  use  of  available  resources   • Boost  communication  among  citizens  and  stakeholders   • Increase  transparency  and  accountability,  as  well  as  equity  and  inclusiveness     • Avoiding  duplication  in  policy  making   • Increasing  responsiveness,  access  and  build  relationship   • Improving  public  image   • Improve  the  quality  of  information   • Consensus  based  decision-­‐making     • Increased  acceptance  of  results     Collaborative  governance  can  be  applied  to  virtually  all  the  policy  making  fields.  The  following  areas   constitute  a  mere  example:   110  |  P a g e  
  • 111.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Infrastructure  management:  building  new  infrastructures  often  entails  the  necessity  to   balance  conflicting  interests,  especially  for  what  concerns  the  case  of  huge  externalities   • Digital   inclusion:   the   increase   in   the   use   of   ICT   has   to   be   fostered   by   the   collaboration   at   every  level   • Energy:  delivering  affordable  and  efficient  energy,  collaboration  on  the  definition  of  energy   regulations   • Environment:   definition   of   new   regulations   on   environmental   safeguard,   mediation   concerning   the   use   of   environmental   resources,   collaboration   in   assessing   public   projects   with  environmental  impact   • Health   care:   collaboration   in   health   care   reform,   awareness   and   education   campaigns,   disease  prevention   • Transportation:   collaboration   in   the   definition   of   a   transportation   plan,   negotiation   of   transportation  rules,  settlement  of  disputes  on  the  construction  of  a  transportation  facility   In   general   the   fields   where   collaborative   governance   is   most   fruitful   are,   in   general,   those   that   (1)   involve  many  different  categories  of  stakeholders,  and  (2)  where  the  conflicts  are  not  frozen  along   entrenched   lines.   Indeed,   (1)   the   benefits   of   collaboration   are   enhanced   when   the   diversity   is   highest,   and   (2)   entrenched   conflicts   are   hardly   solved   by   dialogue,   and   become   a   field   of   power   balance  and  force.       Inspiring  cases  of  ICT  applications  to  Collaborative  Governance     The   Open   Government   Initiative 243  carried   out   by   the   Obama   Administration,   for   promoting   government  transparency  and  citizen  engagement  on  a  global  scale:   • Partner4Solutions244:  the  website  for  the  Partnership  Fund  for  Program  Integrity  Innovation   at   the   Office   of   Management   and   Budget.   By   using   this   tool,   the   Partnership   Fund   aims   at   gathering  ideas  from  citizens  for  improving  the  Federal  assistance  programme     • Regulations.gov245:   in   this   website   it   is   possible   to   comment   on   proposed   regulations   and   related   documents   published   by   the   U.S.   Federal   government,   as   well   as   to   search   and   review  original  regulatory  documents  as  well  as  comments  submitted  by  others   • Challenge.gov246:   online   challenge   platform   allowing   the   public   to   bring   the   best   ideas   and   top   talent   to   bear   on   our   nation’s   most   pressing   challenges,   which   can   range   from   simple   ideas   and   suggestions   to   proofs   of   concept,   designs,   or   finished   products   that   solve   the   grand  challenges  of  the  21st  century   • We   the   People247:   allowing   citizens   to   create   and   launch   a   petition   in   order   to   engage   the   government     • Change  by  Us248,  which  allows  people  to  propose  ideas  and  projects  for  improving  the  cities   they  live.  So  far  the  tool  has  been  applied  to  New  York,  Phoenix  and  Philadelphia   111  |  P a g e  
  • 112.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • Wikirendum249,  a  platform  where  citizens  can  share  ideas  on  policy  making   • The  Digital  Engagement  Guide250,  which  is  a  platform  where  ideas  on  how  to  use  digital  and   social  media  in  the  public  sector  are  shared         Key  challenges  and  gaps     The  key  challenges  and  gaps  in  collaborative  governance  are:   • Coping  with  accelerating  changes  in  policy  making   • Overlapping  in  institutions  and  jurisdictions   • Increasing  complexity  in  the  issues  to  be  tackled   • Ability  to  choose  the  appropriate  tool  for  tackling  the  problem  at  hand   • The  need  to  integrate  policies  and  resources   • Managing  expectations   • Public  involvement  processes  can  be  disconnected  from  real  decision  making   • Tackle  conflicting  interests  among  participants     • Using  tools  appropriate  for  the  scale  (small-­‐scale  or  large  scale)  of  the  problem/solution   • Calibrate  the  level  of  citizens’  participation  required  with  respect  to  the  nature  of  the   problem/solution251   • Define  the  appropriate  levels  of  accountability   • Avoid  instability  in  preferences                                                                                                                                                                                                                                                                                                                                                                                             243   245!home   246   247   248   249   250   251  Making  a  reliable  synthesis  of  a  large-­‐scale  discussion  can  be  daunting.  An  approach  considered  viable  is  some  sort  of   machine-­‐assisted   human   "harvesting",   or   "catching":   software   uses   several   algorithms   to   identify   possible   "atoms   of   interest".  For  example,  networks  analysis  can  detect  balkanization  of  a  community  around  a  polarizing  issue;  if  users  give   ratings  to  statements  (as  is  the  case  in  some  of  the  tools  examined  by  CROSSOVER)  Bayesian  inference  on  user  behavior   can  detect  inconsistencies,  and  therefore  irrational  biases.  The  content  thus  algorithmically  selected  is  then  presented  to   human  "harvesters",  who  can  write  summaries.  These  attention-­‐mediation  algorithms  were  apparently  developed  in  the   context  of  medicine,  as  a  tool  for  helping  doctors  formulate  diagnoses   244 112  |  P a g e  
  • 113.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP         Current  research   • Analysing  the  compatibility  of  new  collaborative  behaviour  with  existing  institutional   framework       Research   disciplines:   political   sciences,   public   administration,   law,   sociology,   and   other   social   sciences   in   general   (including   institutional   economics   for   example),   as   well   as   organisational,   network,  innovation  theories,  etc.   Possible  research  instruments:  thematic  networks,  Support  Action     Future  research:  long  term  and  short  term  issues     Short-­‐term  research   • Updated  institutional  framework     Long-­‐term  research     • New  models  of  governance  and  service  provision       3.2.7. Participatory  Sensing   Introduction  and  definition   Participatory  sensing  refers  to  the  usage  of  sensors,  usually  embedded  in  personal  devices  such  as   smartphones   to   allow   citizens   to   feed   data   of   public   interest.   This   could   include   anything   from   photos   to   passive   monitoring   of   movement   in   the   traffic.   Participatory   sensing   involves   higher   commitment  from  citizens,  contrary  to  opportunistic  sensing  where  user  may  not  be  aware  of  active   applications.   The   diffusion   of   mobile   phones   significantly   lowers   the   barriers   of   participation   and   data  input  by  citizens,  with  automated  geo-­‐tagging  and  time-­‐stamping:  given  the  right  architecture,   they   could   act   as   sensor   nodes   and   location-­‐aware   data   collection   instruments.   While   traditional   sensor  nodes  are  centralised,  these  sensors  are  under  the  owners’  control.  This  would  give  way  to   data  availability  at  an  unprecedented  scale.     Why  it  matters  in  governance   Participatory   sensing   radically   improves   the   data   availability   for   evaluating   the   effect   of   public   policies  and  how  individual  behaviour  is  changing,  provided  adequate  privacy  provisions  are  in  place.   Devices   should   assure   enhanced   users’   control   over   data,   i.e.   which   data   is   being   sent,   when   and   how  it  is  treated,  as  well  as  possibility  for  enhanced  data  anonymisation.   Furthermore,  design  of  participatory  sensing  should  be  placed  in  the  framework  of  policy  contexts,   113  |  P a g e  
  • 114.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   allowing   inference   of   policy   impact   from   data.   Future   platforms   should   combine   participatory   sensing,  mass  moderation,  personalised  feedback  and  social  network  analysis  to  assess  the  interplay   between  perception,  data  and  social  interaction.   Participatory   sensing   is   already   used   in   “public   sphere”   activities   such   as   environment   and   health.   However   the   specific   issue   of   evaluating   public   policies   has   been   so   far   little   researched,   with   particular   regard   to   the   implications   for   privacy,   large-­‐scale   deployment   and   bias   management   on   citizens  sensing.     Recent  trends   Small-­‐scale   experiments   are   being   carried   out   in   different   domains,   mainly   dealing   with   environmental  and  health  data.  Applications  in  the  field  of  urban  planning  are  particularly  promising,   yet   there   is   no   structure   link   between   participatory   sensing   and   policy   models.   Larger   scale   deployment  would  require  more  granular  privacy  compliance  and  user-­‐control,  adequate  incentives   to  participation  and  deriving  business  models.  There  is  no  formalisation  of  the  requirements  and  the   design   of   opportunistic   versus   participatory   sensing,   including   sampling   design   for   participants   recruitment.     Advantage  of  Application  in  Public  Policy     Participatory  sensing  can  be  used  to  gather  and  collect  the  following  kinds  of  information:   • Civic  data:  neighborhood  maintenance  issues,  power  outage  documentation   • Environmental  data:  data  providing  hints  on  pollution  levels,  climate-­‐change  related  data   • Transportation:  commutation  habits,  location  and  movement  data,  condition  of  the  road,   connections  to  public  transportation,  incidence  of  traffic,  accidents  occurrence     • Health:  vital  signs,  info  providing  early  warnings  of  diminishing  health,  info  on  epidemic   spread,  self-­‐administered  diagnostic  tests     The  advantages  for  policy  making  are:   • Possibility  to  collect  data  at  an  otherwise  unachievable  scale  and  geographic  range   • Virtually  costless  data  collection   • Reveal  and  highlight  behavioural  patterns  and  routines  which  can  be  accordingly  changed   • Engage  common  citizens  in  sensitive  issues   • More  pervasive  monitoring  capacity  in  fields  such  as  environment  and  health   Inspiring  cases  of  policy  making  related  applications252   • PEIR 253  (Personal   Environmental   Impact   Report),   which   is   a   system   allowing   users   to   determine  their  exposure  to  environmental  pollution  by  using  a  sensor  in  their  mobile  phone   able  to  determine  the  location  and  the  mean  of  transport   • eHealthSense254,  automatically  detects  health  related  events  which  are  not  directly  observed   by  current  sensor  technology,  like  pain,  tow  conditions,  depression   114  |  P a g e  
  • 115.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   • SenSay255,   which   is   able   to   alert   the   medical   staff   when   the   user   falls   or   in   case   of   suspect   behaviour   • MobAsthma256,  which  monitors  the  exposure  to  pollution  affecting  asthma.  Both  the  volume   of  air  inhaled  and  the  pollution  rate  are  collected  by  sensors  interfaced  to  the  mobile  phone     • Haze   Watch257,   in   which   the   concentration   of   carbon   monoxide,   ozone,   sulphur   dioxide,   and   nitrogen  dioxide  is  measured  by  embedding  pollution  sensors  in  mobile  phones     • NoiseTube258,  which  registers  noise  levels  used  to  monitor  noise  pollution,  which  can  affect   human  hearing  and  behaviour   • EpySurveyor259,  used  by  the  Red  Cross  to  evaluate  anti-­‐malarial  bednet  distribution  and  use   throughout  sub-­‐Saharan  Africa,  as  well  as  the  coverage  achieved  by  vaccination  campaigns   • CarTel260,   which   is   a   mobile   sensing   and   computing   system   making   use   of   mobile   phones   carried  in  vehicles  to  collect  information  about  traffic  or  WIFI  access  points     • NoiseSpy261,  which  is  a  sound-­‐sensing  system  able  to  log  data  for  monitoring  environmental   noise.  Users  can  explore  a  city  area  while  at  the  same  time  visualize  noise  levels  in  real  time     Key  challenges  and  gaps   • Preserve  the  privacy  of  the  users  which  are  required  to  provide  extremely  personal  data   • Create  new  mobile  device  interfaces  which  are  engaging  and  efficient  and  can  be  used   • Ensure  security,  as  the  current  and  past  citizen’s  position  might  be  spotted   • Provide   new   sensors   capable   of   increasing   the   range   of   information   that   individuals   can   track  and  use   • Create  network  infrastructures  aimed  at  supporting  participatory  sensing  services   • Provide  incentive  for  participation  to  data  collection   • Develop   analytical   techniques   to   carry   out   more   accurate   inference   with   mobile   phone   supplied  data  such  as  geo-­‐data  and  images     • Develop   visual   analytics   and   data   analysis   techniques   which   provide   relevant   and   easy   to   interpret  information  for  the  general  public   • Create   engaging   and   efficient   mobile   device   interfaces   to   support   effective,   real-­‐time   user   interaction   • Provide   quality   data,   temporal   and   geo-­‐graphical   availability,   and   ability   to   cover   the   phenomena                                                                                                                                                                                                                                                                                                                                                                                               252  To  the  best  of  our  knowledge  there  are  not  yet  government  applications  in  the  realm  of  participatory  sensing   254   255  See  Siewiorek  et  al.  (2003)   256   257   258   259   260   261   253 115  |  P a g e  
  • 116.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   Current  research   • Aggregating  and  validating  citizens  generated  and  government  data  resource  discovery,     • Selective  sharing,  and  context  verification  mechanisms,  as  well  as  application-­‐level  support   for  data  gathering  campaigns,   • Incentives  for  participatory  sensing,     • Evaluation  of  human  agents  as  sensors   Disciplines   of   research:   sensor   networks,   location   services;   psychology,   economics   of   participation;   privacy   Research  instruments:  testbeds  and  living  labs,  STREPs   Future  research:  long  term  and  short  term  issues   Short-­‐term  research   • Sensing  coverage,  sensor  calibration  and  sensor  context  for  opportunistic  sensing.   • Quality  verification  for  participatory  sensing   • Privacy-­‐compliant  sensing  and  sharing   • Business  models  for  participatory  sensing   • Intelligently  recruiting  collaborators  and  deploying  data  collection  protocols.   • Anonymous,  transparent  use  of  human-­‐carried  sensing  devices   • Evaluating  behavioural  change  through  participatory  sensing   Long-­‐term  research   • •   Enhanced  analytical  techniques  to  make  more  accurate  inferences  from  mobile  phone-­‐ supplied  data  such  as  location  and  images  and  to  automatically  detect  and  respond  to  subtle   events;     New  personal-­‐scale  sensors  to  expand  the  range  of  information  that  individuals  can  track   and  use   • Privacy  by  design  in  participatory  sensing   116  |  P a g e  
  • 117.                                                                                                                              0205F01_INTERNATIONAL  RESEARCH  ROADMAP   3.2.8. Identity  Management       Introduction  and  definition   Digital  identity  management  has  long  been  a  policy  priority  in  the  EU  Member  States,  and  large-­‐scale   investments   have   been   deployed.   In   the   context   of   collaborative   governance,   digital   identity   constitutes  a  fundamental  pillar  of  trustworthy  cooperation.  Identity  management  systems  include   control  and  management  of  credentials  used  to  authenticate  one  entity  to  another,  and  authorise  an   entity  to  adopt  a  specific  role  or  perform  a  specific  task.  Global  in  nature,  they  should  support  non-­‐ repudiation   mechanisms   and   policies;   dynamic   management   of   identities,   roles,   and   permissions;   privacy   protection   mechanisms   and   revocation   of   permissions,   roles,   and   identity   credentials.   Furthermore,   all   the   identities   and   associated   assertions   and   credentials   must   be   machine   processable  and  human  understandable.   At  the  EU  level,  the  goal  is  to  provide  an  interoperable  privacy  protecting  infrastructure  for  eID  that   is   federated   across   countries,   with   multiple   levels   of   security   for   different   services,   relying   on   authentic  sources,  and  usable  in  a  private  sector  context.   Alongside   this,   a   flexible,   context-­‐dependent   and   interoperable   identity   management   system   is   required   for   large-­‐scale   deployment.   In   particular,   federated   identity   management   systems   that   ensure   flexible   deployment   and   seamless   integration   of   users’   preferred   identities,   including   commercial   (such   as   Facebook   connect)   and   open   source   solutions   (such   as   OpenID)   are   needed.   Particular   focus   should   be   put   on   usable   delegation   of   privileges,   which   is   very   important   for   workflows  and  integrating  services.   Electronic  identity  management  should  identify  non-­‐humans  (devices,  sensors)  as  well  as  humans,  in   order  to  ensure  validated  identity  in  the  context  of  participatory  sensing  and  the  Internet  of  Things.   At   the   same   time,   eIdentity   management   should   take   into   account   the   risks   of   information   centralization  in  terms  of  data  privacy  and  security.  Cost-­‐benefit  considerations  of  centralised  versus   federated   systems   remains   a   key   issue.   Identity   federation   can   be   accomplis