SlideShare a Scribd company logo
1 of 21
Download to read offline
                                                                                                                                  	
  
	
  
                                                                                            USING	
  SOCIAL	
  MEDIA	
  AND	
  ONLINE	
  CONVERSATIONS	
  
                                                                                             TO	
  ADD	
  DEPTH	
  TO	
  UNEMPLOYMENT	
  STATISTICS	
  
                                                                                                                                                                                                 Methodological	
  White	
  Paper1	
  
                                                                                                                                                                                                                                                    December	
  8,	
  2011	
  
                                                                                                                                                                                                                                                           	
  
Abstract	
  
This	
   research	
   investigates	
   whether	
   and	
   how	
   social	
   media	
   and	
   other	
   online	
   user-­‐generated	
  
content	
   could	
   enrich	
   understanding	
   of	
   the	
   effect	
   of	
   changing	
   employment	
   conditions.	
   The	
  
primary	
   goal	
   is	
   to	
   compare	
   the	
   qualitative	
   information	
   offered	
   by	
   social	
   media	
   with	
  
unemployment	
   figures.	
   To	
   this	
   end,	
   we	
   first	
   selected	
   online	
   job-­‐related	
   conversations	
   from	
  
blogs,	
   forums	
   and	
   news	
   from	
   the	
   United	
   States	
   and	
   Ireland.	
   For	
   all	
   documents,	
   a	
   quantitative	
  
mood	
   score	
   based	
   on	
   the	
   tone	
   of	
   the	
   conversations—for	
   example	
   happiness,	
   depression	
   or	
  
anxiety—	
  was	
  assigned.	
  The	
  number	
  of	
  unemployment-­‐related	
  documents	
  that	
  also	
  dealt	
  with	
  
other	
  topics	
  such	
  as	
  housing	
  and	
  transportation	
  was	
  also	
  quantified,	
  in	
  order	
  to	
  gain	
  insight	
  into	
  	
  
populations’	
   coping	
   mechanisms.	
   This	
   data	
   was	
   analyzed	
   in	
   two	
   primary	
   ways.	
   First,	
   the	
  
quantified	
   mood	
   scoring	
   was	
   correlated	
   to	
   the	
   unemployment	
   rate	
   to	
   discover	
   leading	
  
indicators	
   that	
   forecast	
   rises	
   and	
   falls	
   in	
   the	
   unemployment	
   rate.	
   For	
   example,	
   the	
   volume	
   of	
  
conversations	
   in	
   Ireland	
   categorized	
   as	
   showing	
   a	
   confused	
   mood	
   correlated	
   with	
   the	
  
unemployment	
   rate	
   with	
   a	
   lead-­‐time	
   of	
   three	
   (3)	
   months.	
   Second,	
   the	
   volume	
   of	
   documents	
  
related	
   to	
   coping	
   mechanisms	
   also	
   showed	
   a	
   significant	
   relationship	
   with	
   the	
   unemployment	
  
rate,	
  which	
  may	
  give	
  insight	
  into	
  the	
  reactions	
  that	
  can	
  be	
  expected	
  from	
  a	
  population	
  dealing	
  
with	
   unemployment.	
   For	
   example,	
   the	
   conversations	
   in	
   the	
   US	
   around	
   the	
   loss	
   of	
   housing	
  
increased	
   two	
   (2)	
   months	
   after	
   unemployment	
   spikes.	
   Overall,	
   in	
   this	
   initial	
   research,	
  SAS	
   and	
  
Global	
   Pulse	
   have	
   underlined	
   the	
   potential	
   of	
   online	
   conversations	
   to	
   complement	
   official	
  
statistics,	
   by	
   providing	
   a	
   qualitative	
   picture	
   demonstrating	
   how	
   people	
   are	
   feeling	
   and	
   coping	
  
with	
  respect	
  to	
  their	
  employment	
  status.	
  

1. Background	
  and	
  Objective	
  
Global	
   Pulse	
   is	
   dedicated	
   to	
   better	
   understanding	
   how	
   new	
   types	
   of	
   data	
   can	
   strengthen	
  
available	
   information	
   on	
   how	
   people	
   are	
   impacted	
   by	
   global	
   crises.	
   This	
   project,	
   conducted	
   in	
  
partnership	
   with	
   SAS,	
   seeks	
   to	
   lay	
   a	
   foundation	
   to	
   use	
   what	
   Global	
   Pulse	
   believes	
   could	
  
represent	
  a	
  powerful	
  source	
  of	
  new	
  data:	
  the	
  global	
  conversation	
  that	
  is	
  taking	
  place	
  over	
  social	
  
media	
  and	
  online	
  content.	
  

	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
   	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
	
  
1
       	
  This methods white paper arose from an on-going series of collaborative research projects conducted by
the United Nations Global Pulse in 2011. Global Pulse is an innovation initiative of the Executive Office of
the UN Secretary-General, which seeks to harness the opportunities in digital data to strengthen evidence-
based decision-making. This research was designed to better understand where digital data can add value
to existing policy analysis, and to contribute to future applications of digital data to global development.
This project was conducted in collaboration with SAS. For more information on this project or the other
projects in this series, please visit: http://www.unglobalpulse.org/research.	
  


White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                                                                                                                                                             1	
     	
     	
  
                                                                                                   	
  
	
  
In	
  particular,	
  this	
  research	
  focuses	
  on	
  what	
  unemployment	
  looks	
  like	
  in	
  social	
  media,	
  specifically	
  
from	
  June	
  2009	
  to	
  June	
  2011.	
  	
  
This	
  project	
  investigates	
  whether	
  and	
  how	
  the	
  feelings	
  and	
  information	
  shared	
  on	
  social	
  media	
  
and	
  other,	
  online,	
  user-­‐generated	
  content	
  could	
  enrich	
  understanding	
  of	
  the	
  effect	
  of	
  changing	
  
employment	
  conditions	
  on	
  people’s	
  perceptions	
  and	
  decisions.	
  	
  
The	
  research	
  zoomed	
  in	
  on	
  the	
  cases	
  of	
  the	
  US	
  and	
  Ireland,	
  where	
  social	
  media	
  and	
  blogs	
  are	
  
widely	
   used	
   and	
   where	
   the	
   recent	
   and	
   ongoing	
   economic	
   crisis	
   has	
   had	
   a	
   severe	
   impact	
   on	
  
employment.	
   In	
   particular,	
   the	
   robust	
   economy	
   and	
   low	
   unemployment	
   in	
   both	
   the	
   US	
   and	
  
Ireland	
  has,	
  over	
  the	
  past	
  two	
  to	
  three	
  years,	
  given	
  way	
  to	
  unprecedented	
  high	
  unemployment	
  
rates	
  (see	
  Figure	
  1).	
  In	
  Ireland,	
  according	
  to	
  EuroStat,	
  the	
  unemployment	
  rate	
  has	
  been	
  under	
  
5%	
   since	
   2000;	
   since	
   the	
   economic	
   crisis,	
   the	
   rate	
   has	
   risen	
   to	
   as	
   high	
   as	
   15%	
   and	
   fluctuated	
  
around	
   14%,	
   a	
   level	
   not	
   seen	
   since	
   the	
   early	
   90s.	
   According	
   to	
   the	
   US	
   Bureau	
   of	
   Labor	
   Statistics,	
  
the	
  unemployment	
  rate	
  in	
  the	
  US	
  reached	
  double	
  digits	
  for	
  the	
  first	
  time	
  since	
  the	
  early	
  80s,	
  and	
  
has	
   dropped	
   slightly	
   from	
   that	
   peak	
   to	
   hover	
   around	
   9%.	
   The	
   organization	
   responsible	
   for	
  
providing	
  the	
  Ireland	
  unemployment	
  rates	
  to	
  the	
  database	
  is	
  the	
  Central	
  Statistics	
  Office,	
  Dublin	
  
and	
  for	
  the	
  United	
  States,	
  the	
  data	
  was	
  provided	
  by	
  the	
  United	
  States	
  Census	
  Bureau.	
  	
  
In	
   this	
   exploratory	
   project,	
   two	
   specific	
   questions	
   were	
   addressed:	
   can	
   online	
   conversations	
  
provide	
   an	
   early	
   indicator	
   of	
   impending	
   job	
   losses,	
   and	
   can	
   they	
   help	
   policy	
   makers	
   enrich	
  
their	
  understanding	
  of	
  the	
  type	
  and	
  sequence	
  of	
  coping	
  strategies	
  employed	
  by	
  individuals?	
  	
  
In	
  order	
  to	
  answer	
  these	
  questions,	
  job-­‐related	
  user-­‐generated	
  unstructured	
  text	
  data	
  was	
  
analyzed	
  using	
  SAS	
  tools	
  and	
  technologies	
  (see	
  Annex	
  A	
  for	
  a	
  detailed	
  list).	
  This	
  allowed	
  
researchers	
  to	
  first	
  retrieve	
  pertinent	
  documents;	
  second,	
  analyze	
  those	
  documents	
  to	
  quantify	
  
the	
  feelings,	
  moods,	
  concerns	
  and	
  strategies;	
  and	
  finally	
  to	
  correlate	
  the	
  data	
  with	
  official	
  
unemployment	
  statistics.	
  
	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                                          2	
     	
     	
  
                                                                                                         	
  
	
  
              a
                      Public	
  Data:
                                                                 Unemployment-­‐
                  Blogs,	
  Forums,	
  News
                                                                     related	
  
                                                                  Conversations
                        US	
  /	
  Ireland




                                       b

                                              Sentiments                            Topics




                                                         c
                                                                  Official	
  
                                                               Unemployment	
  
                                                                 Statistics
                                                                                                	
  
Figure	
  1:	
  Project	
  Workflow	
  

2. Process	
  Overview	
  
The	
  project	
  was	
  implemented	
  according	
  to	
  following	
  workflow	
  (see	
  Figure	
  1):	
  
       a) Online	
   job-­‐related	
   conversations	
   from	
   blogs,	
   forums	
   and	
   news	
   were	
   automatically	
  
          retrieved.	
  
       b) Each	
  document	
  was	
  assigned	
  a	
  quantitative	
  mood	
  score	
  based	
  on	
  the	
  tone	
  or	
  mood	
  of	
  
          the	
   conversations—for	
   example	
   happiness,	
   depression,	
   anxiety—it	
   contained.	
   The	
  
          number	
  of	
  unemployment	
  related	
  documents	
  that	
  also	
  dealt	
  with	
  other	
  topics,	
  such	
  as	
  
          housing	
   and	
   transportation,	
   was	
   quantified	
   and	
   categorized	
   into	
   pre-­‐defined	
   lists	
   of	
  
          document	
  topics	
  representing	
  potential	
  coping	
  mechanisms.	
  
       c) These	
   measures—aggregated	
   mood	
   scores	
   and	
   the	
   volume	
   of	
   conversations	
   around	
  
          different	
   topics—	
   were	
   compared	
   with	
   official	
   unemployment	
   statistics	
   over	
   time	
   in	
  
          search	
  of	
  interesting	
  dynamic-­‐correlations.	
  

3. Data	
  Acquisition	
  and	
  Treatment	
  

3.1.         Acquisition	
  
This	
  process	
  defines	
  the	
  relevant	
  data	
  set.	
  The	
  sample	
  data	
  was	
  pulled	
  from	
  a	
  variety	
  of	
  public	
  
Internet	
  sources,	
  i.e.	
  blogs,	
  internet	
  forums	
  and	
  news,	
  published	
  in	
  the	
  US	
  and	
  Ireland	
  over	
  the	
  
past	
   two	
   years.	
   Relevant	
   documents	
   containing	
   references	
   to	
   terms	
   defining	
   unemployment,	
  



White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                        3	
     	
     	
  
                                                                                                     	
  
	
  
specifically	
   terms	
   and	
   phrases	
   such	
   as	
   “unemployed,	
   fired,	
   on	
   the	
   dole,	
   and	
   collecting	
  
unemployment”,	
   were	
   automatically	
   retrieved	
   and	
   identified.	
   The	
   selection	
   was	
   subsequently	
  
refined	
  using	
  additional	
  unemployment	
  synonyms.	
  

3.2.          Filtering	
  
Once	
  the	
  data	
  set	
  is	
  defined,	
  the	
  documents	
  pass	
  through	
  a	
  custom	
  filtering	
  process	
  meant	
  to	
  
eliminate	
  noise	
  and	
  sort	
  the	
  documents	
  by	
  topic.	
  
The	
  words	
  used	
  in	
  each	
  document	
  were	
  mined	
  in	
  order	
  to	
  assign	
  one	
  or	
  more	
  topical	
  categories.	
  
These	
   categories	
   were	
   chosen	
   to	
   represent	
   some	
   typical	
   coping	
   mechanisms	
   and	
   included:	
  
Housing,	
   Transportation,	
   Entertainment,	
   Consumer	
   Spending,	
   Bills,	
   Financial	
   Distress,	
  
Underemployment,	
   Nutrition	
   Quality,	
   Alcohol,	
   Borrow/Save	
   Money,	
   Education,	
   Healthcare,	
  
Unemployment	
  Claims,	
  and	
  Travel	
  (see	
  Annex	
  B	
  for	
  textual	
  examples	
  of	
  these	
  categories).	
  
Categories	
   related	
   specifically	
   to	
   job	
   status	
   were	
   also	
   constructed,	
   such	
   as	
   the	
   chatter	
   of	
   people	
  
who	
   downgraded	
   by	
   considering	
   employment	
   that	
   paid	
   less	
   or	
   required	
   lower	
   qualifications	
  
than	
   previous	
   employment.	
   Finally,	
   underemployment,	
   where	
   people	
   are	
   working	
   only	
   part	
  
time,	
  was	
  also	
  examined.	
  

3.3.          Validation	
  
Selected	
   documents	
   underwent	
   a	
   series	
   of	
   validation	
   rounds	
   to	
   ensure	
   the	
   quality	
   and	
   accuracy	
  
of	
  the	
  query	
  and	
  filtering	
  processes.	
  For	
  this	
  purpose,	
  a	
  representative	
  sample	
  of	
  the	
  data	
  was	
  
manually	
   reviewed	
   to	
   check	
   the	
   relevance	
   of	
   the	
   content	
   and	
   the	
   adequacy	
   of	
   the	
  
categorization	
  into	
  different	
  topical	
  categories.	
  Once	
  a	
  reasonable	
  number	
  of	
  documents	
  were	
  
successfully	
   reviewed,	
   the	
   query	
   and	
   filtering	
   processes	
   were	
   approved	
   and	
   the	
   resulting	
  
dataset	
  was	
  deemed	
  appropriate	
  for	
  analysis.	
  

4. Analysis	
  

4.1.          Sentiment	
  analysis	
  
Each	
   selected	
   document	
   was	
   then	
   evaluated	
   with	
   sentiment	
   analysis	
   techniques—a	
   form	
   of	
  
automated	
   text	
   analysis	
   that	
   assesses	
   the	
   nature	
   (e.g.	
   anxiety,	
   confusion,	
   hostility,	
   etc.)	
   and	
  
polarity	
   (positive,	
   negative,	
   neutral)	
   of	
   the	
   content	
   expressed	
   using	
   natural	
   language	
   processing	
  
and	
  linguistic	
  rules.	
  Each	
  document	
  was	
  subsequently	
  assigned	
  a	
  mood	
  score,	
   according	
   to	
  six	
  
different	
   sentiments	
   using	
   the	
   SAS’	
   sentiment	
   analysis	
   engine	
   (see	
   Annex	
   B2	
   for	
   textual	
  
examples	
   of	
   these	
   categories).	
   When	
   a	
   word	
   or	
   phrase	
   within	
   a	
   document	
   is	
   a	
   predefined	
  
classifier	
   for	
   one	
   of	
   the	
   mood	
   categories,	
   that	
   document	
   received	
   a	
   numeric	
   score	
   for	
   that	
  
mood.	
  
The	
  application	
  displays	
  the	
  various	
  moods	
  that	
  dominated	
  social	
  media	
  in	
  any	
  given	
  month	
  (as	
  
depicted	
   on	
   the	
   left)	
   or	
   the	
   change	
   in	
   a	
   mood	
   state	
   over	
   time	
   (confidence	
   in	
   the	
   US	
   depicted	
   on	
  
the	
  right).	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                                            4	
     	
     	
  
                                                                                                                      	
  
	
  




                                                                             	
  	
  	
  	
  	
  	
  	
  	
  	
  	
                                                	
  
Figure	
  2:	
  Sentiment	
  Analysis	
  

This	
   analysis	
   gives	
   the	
   user	
   an	
   idea	
   of	
   how	
   the	
   unemployed	
   feel	
   about	
   their	
   status.	
   Are	
   they	
  
optimistic	
   about	
   the	
   future?	
   Are	
   they	
   depressed	
   or	
   anxious	
   about	
   their	
   job	
   prospects?	
   In	
   this	
  
manner,	
   sentiment	
   analysis	
   can	
   provide	
   an	
   important	
   complement	
   to	
   the	
   unemployment	
  
statistics	
  by	
  essentially	
  quantifying	
  the	
  qualitative	
  experience	
  of	
  the	
  unemployed.	
  

4.2.         Cross-­‐Correlation	
  Analysis	
  
Finally,	
  dynamic-­‐correlation	
  tests	
  between	
  mood	
  scores	
  and	
  the	
  volume	
  of	
  documents	
  in	
  a	
  given	
  
category	
   with	
   official	
   unemployment	
   rate	
   were	
   performed	
   in	
   order	
   to	
   reveal,	
   in	
   social	
   media	
  
and	
  online	
  user-­‐generated	
  content:	
  
       •     Leading	
  indicators	
  that	
  could	
  give	
  advance	
  warning	
  about	
  unemployment	
  statistics	
  
       •     Lagging	
   indicators	
   of	
   impact	
   and	
   coping	
   strategies	
   that	
   could	
   be	
   used	
   to	
   design	
   policy	
  
             interventions	
   aimed	
   at	
   mitigating	
   the	
   effects	
   of	
   unemployment,	
   i.e.	
   to	
   assess	
   how	
  
             various	
   types	
   of	
   unemployment	
   chatter	
   might	
   predict	
   the	
   impact	
   of	
   changing	
  
             employment	
   conditions	
   on	
   other	
   public	
   services,	
   such	
   as	
   the	
   use	
   of	
   public	
  
             transportation,	
  housing,	
  etc.	
  
These	
   dynamic-­‐correlation	
   tests	
   were	
   conducted	
   using	
   the	
   maximum	
   value	
   of	
   the	
   average	
  
weekly	
   moods	
   scores	
   assigned	
   to	
   the	
   corpus	
   of	
   selected	
   documents.	
   Coping	
   strategy	
   volumes	
  
are	
  obtained	
  per	
   month.	
  These	
  scores	
  and	
  volumes	
  are	
  tested	
  for	
  dynamic-­‐correlation	
  against	
  
the	
  official	
  unemployment	
  rate.	
  Only	
  results	
  at	
  a	
  90%	
  confidence	
  level	
  or	
  higher	
  are	
  discussed	
  
here	
  and	
  presented	
  in	
  the	
  dashboard.	
  
The	
   category	
   of	
   transportation	
   provides	
   one	
   compelling	
   example.	
   The	
   dynamic-­‐correlation	
  
analysis	
  plot	
  below	
  shows	
  a	
  statistically	
  significant	
  relationship	
  between	
  transportation	
  chatter	
  
and	
  the	
  actual	
  unemployment	
  rate.	
  
	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                                     5	
       	
     	
  
                                                                                               	
  
	
  




                                                                                                                        	
  
Figure	
  3:	
  Cross	
  Series	
  Plot	
  

Upon	
   further	
   investigation,	
   it	
   became	
   clear	
   that	
   discussions	
   about	
   public	
   transportation	
   spike	
  
one	
  month	
  prior	
  to	
  a	
  spike	
  in	
  unemployment.	
  This	
  suggests	
  that	
  an	
  increased	
  demand	
  for	
  public	
  
transportation	
  can	
  be	
  expected	
  to	
  spike	
  prior	
  to	
  a	
  spike	
  in	
  unemployment.	
  




                                                                                                             	
  
Figure	
  4:	
  Cross-­‐Correlations	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                   6	
     	
     	
  
                                                                                                     	
  
	
  
Correlation	
  analysis	
  was	
  conducted	
  for	
  all	
  of	
  the	
  mood	
  states	
  and	
  coping	
  strategies	
  and	
  resulted	
  
in	
  statistically	
  significant	
  correlations	
  between	
  several	
  categories	
  and	
  the	
  official	
  unemployment	
  
rate	
  (see	
  Annex	
  C	
  for	
  more	
  examples	
  of	
  cross-­‐correlation	
  functions).	
  

5. Outcomes	
  

5.1.         Unemployment	
  Dashboard	
  
The	
   SAS	
   team	
   created	
   two	
   Dashboards—one	
   for	
   Ireland	
   and	
   one	
   for	
   the	
   US—	
   that	
   provide	
   a	
  
visual	
   representation	
   of	
   findings.	
   The	
   dashboards	
   were	
   designed	
   to	
   be	
   easily	
   understood	
   by	
  
non-­‐experts,	
  and	
  it	
  is	
  an	
  example	
  of	
  how	
  one	
  might	
  monitor	
  the	
  real	
  time	
  breakdown	
  of	
  social	
  
listening	
  results	
  in	
  the	
  future.	
  
The	
   Dashboard	
   allows	
   the	
   user	
   to	
   investigate	
   the	
   volume	
   of	
   conversations	
   around	
  
unemployment	
   as	
   well	
   as	
   the	
   coping	
   mechanisms	
   that	
   are	
   being	
   discussed	
   in	
   relation	
   to	
  
unemployment.	
   The	
   Dashboard	
   also	
   identifies	
   time	
   relationships	
   between	
   unemployment,	
  
conversation	
   moods,	
   coping	
   mechanisms	
   and	
   various	
   macroeconomic	
   indicators,	
   which	
   allow	
  
the	
  user	
  to	
  see	
  patterns	
  and	
  predictions.	
  Understanding	
  the	
  timing	
  and	
  circumstances	
  that	
  are	
  
causing	
  a	
  shift	
  in	
  a	
  populations’	
  behavior	
  allows	
  decision	
  makers	
  to	
  prepare	
  and	
  act.	
  




                                                                                                                                          	
  
Figure	
  5:	
  United	
  States	
  Dashboard	
  

In	
  the	
  top	
  left	
  quadrant	
  
The	
   official	
   unemployment	
   rate	
   and	
   the	
   forecasting	
   model	
   are	
   depicted.	
   The	
   forecast	
   of	
   the	
  
unemployment	
   rate	
   was	
   obtained	
   using	
   the	
   SAS®	
   Time	
   Series	
   Forecasting	
   System.	
   The	
   model	
  


White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                        7	
     	
     	
  
                                                                                                                         	
  
	
  
with	
   the	
   best	
   Akaike	
   information	
   criterion	
   to	
   measure	
   the	
   relative	
   goodness	
   of	
   fit	
   of	
   a	
   statistical	
  
model	
  is	
  chosen.	
  A	
  Linear	
  Holt	
  Exponential	
  Smoothing	
  model	
  was	
  chosen	
  for	
  the	
  United	
  States	
  
and	
  a	
  Winters	
  Method	
  Additive	
  Model	
  was	
  chosen	
  for	
  Ireland.	
  In	
  the	
  future,	
  the	
  results	
  such	
  as	
  
those	
   from	
   this	
   proof	
   of	
   value	
   study	
   could	
   become	
   valuable	
   inputs	
   for	
   a	
   predictive	
  
unemployment	
  model.	
  This	
  quadrant	
  exemplifies	
  the	
  possible	
  display	
  of	
  such	
  a	
  model.	
  
In	
  the	
  top	
  right	
  quadrant	
  
The	
  volume	
  of	
  documents	
  that	
  discuss	
  particular	
  topics	
  are	
  presented	
  per	
  month	
  in	
  a	
  pie	
  chart.	
  
In	
  the	
  bottom	
  right	
  quadrant	
  
The	
   mood	
   scores	
   per	
   month	
   are	
   shown	
   as	
   six	
   mood	
   dials	
   or	
   barometers.	
   The	
   benefit	
   of	
  
presenting	
  the	
  mood	
  in	
  this	
  manner	
  is	
  that	
  it	
  allows	
  a	
  quick	
  determination	
  of	
  changes	
  over	
  the	
  
past	
  month—in	
  other	
  words;	
  it	
  is	
  easy	
  to	
  see	
  if	
  there	
  has	
  been	
  a	
  negative	
  or	
  positive	
  change	
  in	
  
mood	
  over	
  the	
  past	
  month.	
  This	
  information	
  is	
  also	
  available	
  over	
  time.	
  By	
  clicking	
  on	
  the	
  title	
  
above	
   the	
   mood	
   dials,	
   a	
   new	
   dashboard	
   will	
   load	
   which	
   displays	
   the	
   results	
   for	
   a	
   selected	
   mood	
  
in	
  a	
  bar	
  chart	
  format	
  over	
  time.	
  
In	
  the	
  bottom	
  left	
  quadrant	
  
Dynamic-­‐Correlation	
  results	
  are	
  shown	
  in	
  a	
  vertical	
  bar	
  chart.	
  The	
  time	
  lag	
  of	
  the	
  correlation	
  is	
  
on	
  the	
  x-­‐axis	
  and	
  the	
  variable	
  correlated	
  with	
  the	
  unemployment	
  rate	
  is	
  on	
  the	
  y-­‐axis.	
  
Other	
   features	
   of	
   the	
   dashboard	
   allow	
   the	
   user	
   to	
   understand	
   the	
   analysis	
   better.	
   Additional	
  
links	
  contained	
  in	
  the	
  dashboard	
  will:	
  
      •      provide	
  examples	
  of	
  phrases	
  from	
  actual	
  document	
  which	
  describe	
  the	
  coping	
  strategy	
  
             or	
  mood	
  state	
  
      •      allow	
  the	
  user	
  to	
  explore	
  the	
  dynamic-­‐correlations	
  in	
  more	
  depth	
  
These	
  features	
  allow	
  the	
  user	
  to	
  understand	
  better	
  the	
  nature	
  of	
  the	
  documents	
  and	
  the	
  data	
  
that	
  informs	
  the	
  analysis.	
  
Finally,	
  the	
  dashboard	
  allows	
  users	
  to	
  look	
  at	
  changes	
  over	
  time.	
  Selecting	
  a	
  month	
  changes	
  all	
  
other	
   appropriate	
   visuals,	
   allowing	
   for	
   the	
   exploration	
   of	
   the	
   interplay	
   between	
   the	
  
unemployment	
  rate,	
  the	
  mood	
  of	
  the	
  unemployed	
  and	
  coping	
  mechanisms.	
  The	
  data	
  and	
  results	
  
are	
  represented	
  in	
  a	
  SAS®	
  Business	
  Intelligence	
  Dashboard	
  Interface.	
  

5.2.         Results	
  
Starting	
   from	
   June	
   2009	
   to	
   June	
   2011,	
   the	
   dashboard	
   retrieved	
   over	
   28,000	
   documents	
   for	
  
Ireland	
  and	
  430,000	
  for	
  the	
  United	
  States.	
  Over	
  half	
  of	
  the	
  documents	
  pulled	
  in	
  each	
  contained	
  
coping	
  mechanism	
  classifiers.	
  
The	
   following	
   figures	
   provide	
   a	
   graphical	
  example	
   of	
   the	
   leading	
   and	
   lagging	
   relationships	
   found	
  
in	
  the	
  data:	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                                    8	
     	
     	
  
                                                                                                                  	
  
	
  




                                                                                                                                    	
  
Figure	
  6:	
  United	
  States	
  Unemployment	
  Spike	
  




                                                                                                                             	
  
Figure	
  7:	
  Ireland	
  Unemployment	
  Spike	
  

Overall,	
   among	
   the	
   40+	
   cross-­‐correlations	
   explored,	
   five	
   (5)	
   indicators	
   in	
   the	
   US	
   and	
   six	
   (6)	
  
indicators	
  in	
  Ireland	
  showed	
  a	
  correlation	
  significant	
  at	
  the	
  90%	
  confidence	
  level	
  or	
  higher.	
  	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                                9	
     	
     	
  
                                                                                                               	
  
	
  
These	
  findings	
  are	
  a	
  first	
  step	
  in	
  this	
  kind	
  of	
  research	
  and	
  represent	
  a	
  high	
  potential	
  source	
  of	
  
complementary	
   information	
   to	
   the	
   unemployment	
   statistics.	
   The	
   following	
   images	
   depict	
   the	
  
strongest	
  dynamic-­‐correlations	
  that	
  were	
  found:	
  
	
  




                                                                                                                              	
  
Figure	
  8:	
  United	
  States	
  Chatter	
  




                                                                                                                              	
  
Figure	
  9:	
  Ireland	
  Chatter	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                        10	
     	
     	
  
                                                                                                                 	
  
	
  
	
  

Table	
  1:	
  Country	
  Correlation,	
  CCF	
  and	
  Significance	
  Level	
  

Country	
              Correlation	
  Description	
                                             CCF	
                            Significance	
  
                                                                                                                                 Level	
  
US	
                   Hostile	
  Mood	
  increases	
  4	
  months	
  before	
  a	
             .442	
                           95%	
  
                       spike	
  in	
  unemployment	
  
US	
                   Depressed	
  Mood	
  increases	
  4	
  months	
                          .489	
                           95%	
  
                       before	
  a	
  spike	
  in	
  unemployment	
  
US	
                   Uncertain	
  Mood	
  increases	
  as	
                                   .448	
                           95%	
  
                       unemployment	
  spikes	
  
US	
                   Talk	
  about	
  loss	
  of	
  housing	
  increases	
  2	
               .634	
                           95%	
  
                       months	
  after	
  an	
  unemployment	
  spike	
  
US	
                   Talk	
  about	
  auto	
  repossession	
  increases	
  3	
                .455	
                           95%	
  
                       months	
  after	
  an	
  unemployment	
  spike	
  
IRELAND	
              Anxious	
  Mood	
  increases	
  5	
  months	
  before	
                  .387	
                           90%	
  
                       a	
  spike	
  in	
  unemployment	
  
IRELAND	
              Confused	
  Mood	
  increases	
  3	
  months	
                           .675	
                           95%	
  
                       before	
  a	
  spike	
  in	
  unemployment	
  
IRELAND	
              Confident	
  Mood	
  decreases	
  2	
  months	
                          -­‐.407	
                        90%	
  
                       before	
  a	
  spike	
  in	
  unemployment	
  
IRELAND	
              Talk	
  about	
  changing	
  transportation	
                            .380	
                           90%	
  
                       methods	
  for	
  the	
  worse	
  increases	
  as	
  
                       unemployment	
  increases	
  
IRELAND	
              Talk	
  about	
  travel	
  cancelations	
  increases	
  3	
              .450	
                           95%	
  
                       months	
  after	
  an	
  unemployment	
  spike	
  
IRELAND	
              Talk	
  about	
  changing	
  housing	
  situations	
  for	
              .328	
                           90%	
  
                       the	
  worse	
  increases	
  8	
  months	
  after	
  
                       unemployment	
  increases	
  

6. Conclusions	
  and	
  perspectives	
  
In	
   this	
   initial	
   research,	
   the	
   high	
   potential	
   of	
   online	
   conversations	
   to	
   complement	
   official	
  
statistics	
  has	
  been	
  shown,	
  by	
  providing	
  leading	
  and	
  lagging	
  indicators	
  that	
  show	
  how	
  people	
  are	
  
feeling	
   with	
   respect	
   to	
   their	
   employment	
   status	
   and	
   which	
   coping	
   strategies	
   (conversation	
  
topics)	
  are	
  employed	
  over	
  time.	
  
Several	
  mood	
  score	
  volumes	
  produced	
  strong	
  correlations	
  with	
  the	
  unemployment	
  rate,	
  which	
  
may	
   be	
   leading	
   indicators	
   that	
   the	
   unemployment	
   will	
   rise	
   or	
   fall.	
   For	
   example,	
   conversations	
   in	
  
Ireland	
   showing	
   a	
   confused	
   mood	
   preceded	
   the	
   unemployment	
   rate	
   variations	
   by	
   3	
   months.	
  
And	
   data	
   pulled	
   from	
   the	
   social	
   listening	
   sources,	
   representing	
   daily	
   conversations,	
   proved	
   to	
  
contain	
  valuable	
  information	
  related	
  to	
  how	
  the	
  unemployed	
  cope.	
  In	
  addition,	
  changes	
  in	
  the	
  
volume	
   of	
   particular	
   coping	
   topics	
   showed	
   significant	
   lagging	
   relationships	
   with	
   the	
  
unemployment	
  rate	
  that	
  may	
  give	
  insight	
  into	
  reactions	
  that	
  can	
  be	
  expected	
  from	
  a	
  population	
  
dealing	
  with	
  unemployment	
  –	
  as	
  demonstrated,	
  for	
  example,	
  by	
  the	
  increase	
  in	
  conversations	
  in	
  
the	
  US	
  around	
  the	
  loss	
  of	
  housing	
  2	
  months	
  after	
  an	
  increase	
  in	
  unemployment.	
  	
  



White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                                 11	
     	
     	
  
                                                                                                                	
  
	
  
In	
  short,	
  the	
  type	
  of	
  data	
  needed	
  to	
  perform	
  this	
  sort	
  of	
  analysis	
  exists,	
  and	
  the	
  initial	
  research	
  
pointed	
   to	
   potential	
   indicators	
   that	
   might	
   be	
   used	
   for	
   improving	
   unemployment	
   policies	
   or	
  
social	
  protection	
  programs.	
  
Several	
   challenges	
   in	
   this	
   type	
   of	
   analysis	
   and	
   future	
   lines	
   of	
   research	
   were	
   also	
   made	
   clear	
  
through	
   the	
   course	
   of	
   this	
   work.	
   First,	
   greater	
   geographical	
   information	
   related	
   to	
   each	
  
document	
  would	
  allow	
  for	
  a	
  finer	
  grain	
  analysis	
  at	
  a	
  regional	
  level.	
  In	
  addition,	
  one	
  of	
  the	
  main	
  
challenges	
  of	
  any	
  social	
  media	
  analysis	
  is	
  to	
  understand	
  better	
  the	
  demographic	
  characteristics	
  
of	
  the	
  sampled	
  population.	
  Finally,	
  online	
  conversations	
  are	
  a	
  relatively	
  new	
  source	
  of	
  data,	
  and	
  
it	
   is	
   thus	
   required	
   to	
   start	
   maintaining	
   a	
   database	
   of	
   these	
   documents	
   over	
   time,	
   so	
   in	
   the	
  
future,	
  analysis	
  will	
  cover	
  a	
  longer	
  period.	
  
The	
   potential	
   is	
   high,	
   and	
   in	
   the	
   future,	
   models	
   that	
   are	
   more	
   robust	
   can	
   be	
   built	
   for	
   both	
  
helping	
  predict	
  and	
  anticipate	
  the	
  consequences	
  of	
  unemployment	
  spikes,	
  and	
  real-­‐time	
  social	
  
listening	
   could	
   be	
   used	
   to	
   monitor	
   the	
   experience	
   of	
   the	
   unemployment	
   in	
   an	
   automated	
  
manner,	
  continuously.	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                                  12	
     	
     	
  
                                                                                                                 	
  
	
  

Annex	
  A:	
  Tools	
  
Data	
  acquisition:	
  The	
  data	
  acquisition	
  is	
  handled	
  through	
  a	
  third	
  party	
  service,	
  Boardreader.	
  
Data	
   filtering:	
   SAS®	
   Content	
   Categorization	
   software	
   is	
   used	
   to	
   create	
   the	
   filter	
   the	
   relevant	
  
documents.	
  Data	
  must	
  satisfy	
  the	
  standards	
  set	
  forth	
  in	
  SAS	
  Content	
  Categorization	
  to	
  be	
  passed	
  
to	
  the	
  next	
  step	
  in	
  the	
  process.	
  
Dynamic-­‐Correlation	
   analysis:	
   Several	
   tools	
   have	
   been	
   used	
   for	
   the	
   analysis,	
   including	
   SAS®	
  
Content	
  Categorization	
  Studio,	
  SAS®	
  9.2	
  statistical	
  and	
  forecasting	
  tools,	
  SAS®	
  Enterprise	
  Guide	
  
and	
   SAS®	
   Sentiment	
   Studio	
   Workbench.	
   The	
   text	
   is	
   parsed	
   through	
   a	
   SAS	
   procedure	
   called	
  
DOCPARSE.	
   The	
   SAS	
   procedure,	
   PROC	
   TIMESERIES,	
   is	
   used	
   to	
   obtain	
   the	
   cross-­‐correlation	
  
function	
  results.	
  
Mood	
   State:	
   Mood	
   State	
   is	
   a	
   method	
   by	
   which	
   SAS	
   measures	
   the	
   overall	
   mood	
   and	
   specific	
  
moods	
  of	
  a	
  data	
  corpus.	
  Unlike	
  sentiment	
  analysis,	
  which	
  is	
  a	
  simple	
  positive/negative/neutral	
  
decision,	
   mood	
   state	
   analysis	
   offers	
   a	
   more	
   refined	
   measure	
   by	
   which	
   to	
   judge	
   social	
   media.	
  
Documents	
   are	
   scored	
   to	
   provide	
   mood	
   scores	
   for	
   Anxiety,	
   Confidence,	
   Hostility,	
   Confusion,	
  
Energy,	
  and	
  Happiness.	
  
Unemployment	
   dashboard:	
   The	
   data	
   and	
   results	
   are	
   represented	
   using	
   the	
   SAS	
   Business	
  
Intelligence	
  Dashboard	
  Interface.	
  
	
  
	
  
	
  
About	
  SAS	
  
SAS	
   is	
   the	
   leader	
   in	
   business	
   analytics	
   software	
   and	
   services,	
   and	
   the	
   largest	
   independent	
  
vendor	
  in	
  the	
  business	
  intelligence	
  market.	
  Through	
  innovative	
  solutions,	
  SAS	
  helps	
  customers	
  
at	
   more	
   than	
   50,000	
   sites	
   improve	
   performance	
   and	
   deliver	
   value	
   by	
   making	
   better	
   decisions	
  
faster.	
  Since	
  1976	
  SAS	
  has	
  been	
  giving	
  customers	
  around	
  the	
  world	
  THE	
  POWER	
  TO	
  KNOW®.	
  SAS	
  
and	
   all	
   other	
   SAS	
   Institute	
   Inc.	
   product	
   or	
   service	
   names	
   are	
   registered	
   trademarks	
   or	
  
trademarks	
   of	
   SAS	
   Institute	
   Inc.	
   in	
   the	
   USA	
   and	
   other	
   countries.	
   ®	
   indicates	
   USA	
   registration.	
  
Other	
   brand	
   and	
   product	
   names	
   are	
   trademarks	
   of	
   their	
   respective	
   companies.	
   Copyright	
   ©	
  
2011	
  SAS	
  Institute	
  Inc.	
  All	
  rights	
  reserved.	
  
	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                               13	
     	
     	
  
                                                                                                              	
  
	
  

Annex	
  B	
  
Example	
  of	
  documents	
  categorization	
  based	
  on	
  conversation	
  topics	
  
Topic	
  Examples	
  (all	
  occur	
  within	
  same	
  document	
  as	
  mention	
  of	
  Unemployment	
  or	
  synonym)	
  
       •     Alcohol	
  
                  o “My	
  beer	
  budget	
  will	
  obviously	
  be	
  cut.”	
  
       •     Borrow/Save	
  money	
  
                  o “I	
  hate	
  to	
  do	
  it	
  but	
  will	
  have	
  to	
  get	
  a	
  payday	
  loan.”	
  
       •     Consumer	
  goods	
  
                  o “Definitely	
  won’t	
  be	
  buying	
  new	
  clothes	
  this	
  season.”	
  
       •     Nutrition	
  quality	
  
                  o “Not	
  sure	
  how	
  to	
  afford	
  fresh	
  produce	
  these	
  days.”	
  
       •     Education	
  
                  o “We	
  can’t	
  afford	
  the	
  tuition	
  so	
  she’ll	
  have	
  to	
  withdraw.”	
  
       •     Underemployment	
  
                  o “It’s	
  only	
  part	
  time	
  but	
  at	
  least	
  it’s	
  money	
  coming	
  in.”	
  
       •     Entertainment	
  
                  o “The	
  kids	
  will	
  make	
  do	
  without	
  new	
  video	
  games	
  or	
  dvds.”	
  
       •     Financial	
  
                  o “A	
  few	
  more	
  months	
  and	
  we’ll	
  have	
  to	
  seriously	
  consider	
  a	
  bankruptcy.”	
  
       •     Bills	
  
                  o “Sorry	
  water	
  bill,	
  this	
  month	
  I	
  pay	
  the	
  electric,	
  next	
  month	
  it’s	
  the	
  student	
  
                         loans.”	
  
       •     Healthcare	
  
                  o “He	
  has	
  cobra	
  available	
  but	
  I	
  don’t	
  know	
  we’ll	
  ever	
  pay	
  for	
  it.”	
  
       •     Housing	
  
                  o Loss	
  
                             § “That’s	
  it.	
  The	
  bank	
  is	
  foreclosing	
  next	
  week.”	
  
                  o Downgrade	
  
                             § “We’ll	
  save	
  a	
  bunch	
  if	
  we	
  can	
  move	
  to	
  a	
  smaller	
  place.”	
  
       •     Transportation	
  
                  o Loss	
  
                             § “They’re	
  trying	
  to	
  repo	
  my	
  wife’s	
  car.”	
  
                  o Downgrade	
  
                             § “I	
  sold	
  the	
  Audi	
  and	
  got	
  a	
  used	
  Mazda.”	
  
                  o Public	
  
                             § “I	
  canceled	
  the	
  car	
  insurance	
  so	
  I’ll	
  start	
  taking	
  the	
  bus.	
  It’s	
  cheaper.”	
  
       •     Travel	
  



White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                            14	
     	
     	
  
                                                                                                        	
  
	
  
                o “The	
  Easter	
  vacation	
  we	
  planned	
  is	
  officially	
  canceled	
  due	
  to	
  lack	
  of	
  funds.”	
  
       •     Unemployment	
  claim	
  
                o “I	
  filed	
  for	
  benefits	
  the	
  same	
  day.	
  Ughh.”	
  

Example	
  of	
  documents	
  categorization	
  based	
  on	
  sentiment	
  analysis	
  
       •     Anxiety	
  
                o “I’m	
  nervous	
  that	
  I	
  won’t	
  find	
  another	
  job	
  like	
  this	
  one.”	
  
       •     Confidence	
  
                o “She	
  doubts	
  that	
  we	
  can	
  afford	
  to	
  have	
  two	
  cars.”	
  
       •     Hostility	
  
                o “It’s	
  outrageous	
  that	
  we	
  were	
  only	
  given	
  two	
  weeks	
  severance!”	
  
       •     Confusion	
  
                o “I’ve	
  felt	
  so	
  scattered	
  since	
  the	
  layoffs.”	
  
       •     Energy	
  
                o “My	
  husband	
  has	
  been	
  so	
  sluggish	
  for	
  the	
  past	
  few	
  months.”	
  
       •     Happiness	
  
                o “The	
  counselor	
  said	
  it	
  was	
  normal	
  to	
  experience	
  a	
  sense	
  of	
  mourning	
  after	
  
                        losing	
  a	
  job.”	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                                      15	
     	
     	
  
                                                                                        	
  
	
  
Annex	
  C	
  
Cross-­‐correlations	
  functions	
  between	
  mood	
  states	
  and/or	
  conversation	
  topics	
  against	
  the	
  
official	
  unemployment	
  rate.	
  

United	
  States	
  




                                                                                                         	
  
Figure	
  10:	
  (Hostile	
  Mood	
  –	
  Agreeable	
  Mood)	
  Hostility	
  




                                                                                                  	
  
Figure	
  11:	
  (Depressed	
  Mood	
  –	
  Elated	
  Mood)	
  Depression	
  



White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                                       16	
     	
     	
  
                                                                         	
  
	
  




                                                                                                	
  
Figure	
  12:	
  (Unsure	
  Mood	
  –	
  Confident	
  Mood)	
  Uncertainty	
  




                                                                                                       	
  
Figure	
  13:	
  (Housing	
  Loss)	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                          17	
     	
     	
  
                                                                            	
  
	
  




                                                                                                       	
  
Figure	
  14:	
  (Transportation	
  Repossession)	
  

Ireland	
  




                                                                                                	
  
Figure	
  15:	
  (Anxious	
  –	
  Composed	
  News)	
  Anxiety	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                          18	
     	
     	
  
                                                                         	
  
	
  




                                                                                                       	
  
Figure	
  16:	
  (Confused	
  –	
  Clearheaded)	
  Confusion	
  




                                                                                                	
  
Figure	
  17:	
  Confident	
  by	
  Blogs	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                          19	
     	
     	
  
                                                                       	
  
	
  




                                                                                                   	
  
Figure	
  18:	
  Transportation	
  Downgrade	
  




                                                                                                	
  
Figure	
  19:	
  Travel	
  Cancellations	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                      20	
     	
     	
  
                                                                     	
  
	
  




                                                                                                	
  
Figure	
  20:	
  Housing	
  Downgrade	
  




White	
  Paper:	
  “Using	
  Social	
  Media	
  to	
  Add	
  Depth	
  to	
  Unemployment”	
                   21	
     	
     	
  

More Related Content

What's hot

Using Twitter to Understand the Post-2015 Global Conversation - Project Overview
Using Twitter to Understand the Post-2015 Global Conversation - Project OverviewUsing Twitter to Understand the Post-2015 Global Conversation - Project Overview
Using Twitter to Understand the Post-2015 Global Conversation - Project OverviewUN Global Pulse
 
Using Twitter Data to Analyse Public Sentiment on Fuel Subsidy Policy Reform ...
Using Twitter Data to Analyse Public Sentiment on Fuel Subsidy Policy Reform ...Using Twitter Data to Analyse Public Sentiment on Fuel Subsidy Policy Reform ...
Using Twitter Data to Analyse Public Sentiment on Fuel Subsidy Policy Reform ...UN Global Pulse
 
Mining Citizen Feedback Data for Enhanced Local Government Decision-Making - ...
Mining Citizen Feedback Data for Enhanced Local Government Decision-Making - ...Mining Citizen Feedback Data for Enhanced Local Government Decision-Making - ...
Mining Citizen Feedback Data for Enhanced Local Government Decision-Making - ...UN Global Pulse
 
Analyzing Attitudes Towards Biofuels with Social Media - Project Overview
Analyzing Attitudes Towards Biofuels with Social Media - Project OverviewAnalyzing Attitudes Towards Biofuels with Social Media - Project Overview
Analyzing Attitudes Towards Biofuels with Social Media - Project OverviewUN Global Pulse
 
Nowcasting Food Prices in Indonesia with Social Media - Project Overview
Nowcasting Food Prices in Indonesia with Social Media - Project Overview  Nowcasting Food Prices in Indonesia with Social Media - Project Overview
Nowcasting Food Prices in Indonesia with Social Media - Project Overview UN Global Pulse
 
Crowdsourcing High- Frequency Food Price Data in Rural Indonesia - Project Ov...
Crowdsourcing High- Frequency Food Price Data in Rural Indonesia - Project Ov...Crowdsourcing High- Frequency Food Price Data in Rural Indonesia - Project Ov...
Crowdsourcing High- Frequency Food Price Data in Rural Indonesia - Project Ov...UN Global Pulse
 
Analyzing Attitudes Towards Contraception & Teenage Pregnancy Using Social Da...
Analyzing Attitudes Towards Contraception & Teenage Pregnancy Using Social Da...Analyzing Attitudes Towards Contraception & Teenage Pregnancy Using Social Da...
Analyzing Attitudes Towards Contraception & Teenage Pregnancy Using Social Da...UN Global Pulse
 
PSFK Future Of Real-Time Information
PSFK Future Of Real-Time InformationPSFK Future Of Real-Time Information
PSFK Future Of Real-Time InformationPSFK
 
Proceedings from International Conference on Data Innovation For Policy Makers
Proceedings from International Conference on Data Innovation For Policy MakersProceedings from International Conference on Data Innovation For Policy Makers
Proceedings from International Conference on Data Innovation For Policy MakersUN Global Pulse
 
Data privacy and security in ICT4D - Meeting Report
Data privacy and security in ICT4D - Meeting Report Data privacy and security in ICT4D - Meeting Report
Data privacy and security in ICT4D - Meeting Report UN Global Pulse
 
Reuters digital news report 2013
Reuters digital news report 2013Reuters digital news report 2013
Reuters digital news report 2013Xosé María Cid
 
Chung-Jui LAI - Polarization of Political Opinion by News Media
Chung-Jui LAI - Polarization of Political Opinion by News MediaChung-Jui LAI - Polarization of Political Opinion by News Media
Chung-Jui LAI - Polarization of Political Opinion by News MediaREVULN
 
Designing social media policy for government: Eight essential elements
Designing social media policy for government: Eight essential elementsDesigning social media policy for government: Eight essential elements
Designing social media policy for government: Eight essential elementsThe Collaboration Project
 
Big Data For Development A Primer
Big Data For Development A PrimerBig Data For Development A Primer
Big Data For Development A PrimerUN Global Pulse
 
Hate Speech and Nigeria’s Struggle for Democratic Consolidation: A Conceptual...
Hate Speech and Nigeria’s Struggle for Democratic Consolidation: A Conceptual...Hate Speech and Nigeria’s Struggle for Democratic Consolidation: A Conceptual...
Hate Speech and Nigeria’s Struggle for Democratic Consolidation: A Conceptual...Premier Publishers
 
El periodismo de datos en 2017
El periodismo de datos en 2017El periodismo de datos en 2017
El periodismo de datos en 2017María Rubio
 
Reuters Institute Digital News Report 2019 final
Reuters Institute Digital News Report 2019 finalReuters Institute Digital News Report 2019 final
Reuters Institute Digital News Report 2019 finalSABC News
 

What's hot (20)

Using Twitter to Understand the Post-2015 Global Conversation - Project Overview
Using Twitter to Understand the Post-2015 Global Conversation - Project OverviewUsing Twitter to Understand the Post-2015 Global Conversation - Project Overview
Using Twitter to Understand the Post-2015 Global Conversation - Project Overview
 
Using Twitter Data to Analyse Public Sentiment on Fuel Subsidy Policy Reform ...
Using Twitter Data to Analyse Public Sentiment on Fuel Subsidy Policy Reform ...Using Twitter Data to Analyse Public Sentiment on Fuel Subsidy Policy Reform ...
Using Twitter Data to Analyse Public Sentiment on Fuel Subsidy Policy Reform ...
 
Mining Citizen Feedback Data for Enhanced Local Government Decision-Making - ...
Mining Citizen Feedback Data for Enhanced Local Government Decision-Making - ...Mining Citizen Feedback Data for Enhanced Local Government Decision-Making - ...
Mining Citizen Feedback Data for Enhanced Local Government Decision-Making - ...
 
Analyzing Attitudes Towards Biofuels with Social Media - Project Overview
Analyzing Attitudes Towards Biofuels with Social Media - Project OverviewAnalyzing Attitudes Towards Biofuels with Social Media - Project Overview
Analyzing Attitudes Towards Biofuels with Social Media - Project Overview
 
Nowcasting Food Prices in Indonesia with Social Media - Project Overview
Nowcasting Food Prices in Indonesia with Social Media - Project Overview  Nowcasting Food Prices in Indonesia with Social Media - Project Overview
Nowcasting Food Prices in Indonesia with Social Media - Project Overview
 
Crowdsourcing High- Frequency Food Price Data in Rural Indonesia - Project Ov...
Crowdsourcing High- Frequency Food Price Data in Rural Indonesia - Project Ov...Crowdsourcing High- Frequency Food Price Data in Rural Indonesia - Project Ov...
Crowdsourcing High- Frequency Food Price Data in Rural Indonesia - Project Ov...
 
Analyzing Attitudes Towards Contraception & Teenage Pregnancy Using Social Da...
Analyzing Attitudes Towards Contraception & Teenage Pregnancy Using Social Da...Analyzing Attitudes Towards Contraception & Teenage Pregnancy Using Social Da...
Analyzing Attitudes Towards Contraception & Teenage Pregnancy Using Social Da...
 
PSFK Future Of Real-Time Information
PSFK Future Of Real-Time InformationPSFK Future Of Real-Time Information
PSFK Future Of Real-Time Information
 
Proceedings from International Conference on Data Innovation For Policy Makers
Proceedings from International Conference on Data Innovation For Policy MakersProceedings from International Conference on Data Innovation For Policy Makers
Proceedings from International Conference on Data Innovation For Policy Makers
 
Mariusz Luterek: E-government as a research field
Mariusz Luterek: E-government as a research field Mariusz Luterek: E-government as a research field
Mariusz Luterek: E-government as a research field
 
Data privacy and security in ICT4D - Meeting Report
Data privacy and security in ICT4D - Meeting Report Data privacy and security in ICT4D - Meeting Report
Data privacy and security in ICT4D - Meeting Report
 
Reuters digital news report 2013
Reuters digital news report 2013Reuters digital news report 2013
Reuters digital news report 2013
 
Chung-Jui LAI - Polarization of Political Opinion by News Media
Chung-Jui LAI - Polarization of Political Opinion by News MediaChung-Jui LAI - Polarization of Political Opinion by News Media
Chung-Jui LAI - Polarization of Political Opinion by News Media
 
Designing social media policy for government: Eight essential elements
Designing social media policy for government: Eight essential elementsDesigning social media policy for government: Eight essential elements
Designing social media policy for government: Eight essential elements
 
Big Data For Development A Primer
Big Data For Development A PrimerBig Data For Development A Primer
Big Data For Development A Primer
 
Estudio data journalism-in-2017 - google
Estudio data journalism-in-2017 - googleEstudio data journalism-in-2017 - google
Estudio data journalism-in-2017 - google
 
Hate Speech and Nigeria’s Struggle for Democratic Consolidation: A Conceptual...
Hate Speech and Nigeria’s Struggle for Democratic Consolidation: A Conceptual...Hate Speech and Nigeria’s Struggle for Democratic Consolidation: A Conceptual...
Hate Speech and Nigeria’s Struggle for Democratic Consolidation: A Conceptual...
 
El periodismo de datos en 2017
El periodismo de datos en 2017El periodismo de datos en 2017
El periodismo de datos en 2017
 
Reuters Institute Digital News Report 2019 final
Reuters Institute Digital News Report 2019 finalReuters Institute Digital News Report 2019 final
Reuters Institute Digital News Report 2019 final
 
Digital News Report 2019
Digital News Report 2019Digital News Report 2019
Digital News Report 2019
 

Viewers also liked

A Guide to Data Innovation for Development - From idea to proof-of-concept
A Guide to Data Innovation for Development - From idea to proof-of-conceptA Guide to Data Innovation for Development - From idea to proof-of-concept
A Guide to Data Innovation for Development - From idea to proof-of-conceptUN Global Pulse
 
Estimating Migration Flows Using Online Search Data - Project Overview
Estimating Migration Flows Using Online Search Data - Project Overview Estimating Migration Flows Using Online Search Data - Project Overview
Estimating Migration Flows Using Online Search Data - Project Overview UN Global Pulse
 
Integrating big data into the monitoring and evaluation of development progra...
Integrating big data into the monitoring and evaluation of development progra...Integrating big data into the monitoring and evaluation of development progra...
Integrating big data into the monitoring and evaluation of development progra...UN Global Pulse
 
MED Watch Dashboard - Big Data Mining on Safety Information and Adverse Event...
MED Watch Dashboard - Big Data Mining on Safety Information and Adverse Event...MED Watch Dashboard - Big Data Mining on Safety Information and Adverse Event...
MED Watch Dashboard - Big Data Mining on Safety Information and Adverse Event...László Köleséri
 
Smugglers and vulnerable migrants in central america and mexico final
Smugglers and vulnerable migrants in central america and mexico   finalSmugglers and vulnerable migrants in central america and mexico   final
Smugglers and vulnerable migrants in central america and mexico finalUN Global Pulse
 
Supporting Forest and Peat Fire Management Using Social Media - Project Overview
Supporting Forest and Peat Fire Management Using Social Media - Project OverviewSupporting Forest and Peat Fire Management Using Social Media - Project Overview
Supporting Forest and Peat Fire Management Using Social Media - Project OverviewUN Global Pulse
 
The impact of the global financial crisis on reproductive and maternal health...
The impact of the global financial crisis on reproductive and maternal health...The impact of the global financial crisis on reproductive and maternal health...
The impact of the global financial crisis on reproductive and maternal health...UN Global Pulse
 
Study on the human dimensions of the financial crisis in ethiopia final
Study on the human dimensions of the financial crisis in ethiopia finalStudy on the human dimensions of the financial crisis in ethiopia final
Study on the human dimensions of the financial crisis in ethiopia finalUN Global Pulse
 
The global financial crisis in colombia and the international conference on p...
The global financial crisis in colombia and the international conference on p...The global financial crisis in colombia and the international conference on p...
The global financial crisis in colombia and the international conference on p...UN Global Pulse
 
Monitoring the implications of the global financial crisis on primary schools...
Monitoring the implications of the global financial crisis on primary schools...Monitoring the implications of the global financial crisis on primary schools...
Monitoring the implications of the global financial crisis on primary schools...UN Global Pulse
 
A visual analytics approach to understanding poverty assessment through disas...
A visual analytics approach to understanding poverty assessment through disas...A visual analytics approach to understanding poverty assessment through disas...
A visual analytics approach to understanding poverty assessment through disas...UN Global Pulse
 
Supporting the Post-2015 Development Agenda Consultations Using U-Report - Pr...
Supporting the Post-2015 Development Agenda Consultations Using U-Report - Pr...Supporting the Post-2015 Development Agenda Consultations Using U-Report - Pr...
Supporting the Post-2015 Development Agenda Consultations Using U-Report - Pr...UN Global Pulse
 
Mobile Data for Development Primer
Mobile Data for Development PrimerMobile Data for Development Primer
Mobile Data for Development PrimerUN Global Pulse
 

Viewers also liked (14)

A Guide to Data Innovation for Development - From idea to proof-of-concept
A Guide to Data Innovation for Development - From idea to proof-of-conceptA Guide to Data Innovation for Development - From idea to proof-of-concept
A Guide to Data Innovation for Development - From idea to proof-of-concept
 
Estimating Migration Flows Using Online Search Data - Project Overview
Estimating Migration Flows Using Online Search Data - Project Overview Estimating Migration Flows Using Online Search Data - Project Overview
Estimating Migration Flows Using Online Search Data - Project Overview
 
Integrating big data into the monitoring and evaluation of development progra...
Integrating big data into the monitoring and evaluation of development progra...Integrating big data into the monitoring and evaluation of development progra...
Integrating big data into the monitoring and evaluation of development progra...
 
MED Watch Dashboard - Big Data Mining on Safety Information and Adverse Event...
MED Watch Dashboard - Big Data Mining on Safety Information and Adverse Event...MED Watch Dashboard - Big Data Mining on Safety Information and Adverse Event...
MED Watch Dashboard - Big Data Mining on Safety Information and Adverse Event...
 
Smugglers and vulnerable migrants in central america and mexico final
Smugglers and vulnerable migrants in central america and mexico   finalSmugglers and vulnerable migrants in central america and mexico   final
Smugglers and vulnerable migrants in central america and mexico final
 
Supporting Forest and Peat Fire Management Using Social Media - Project Overview
Supporting Forest and Peat Fire Management Using Social Media - Project OverviewSupporting Forest and Peat Fire Management Using Social Media - Project Overview
Supporting Forest and Peat Fire Management Using Social Media - Project Overview
 
The impact of the global financial crisis on reproductive and maternal health...
The impact of the global financial crisis on reproductive and maternal health...The impact of the global financial crisis on reproductive and maternal health...
The impact of the global financial crisis on reproductive and maternal health...
 
Study on the human dimensions of the financial crisis in ethiopia final
Study on the human dimensions of the financial crisis in ethiopia finalStudy on the human dimensions of the financial crisis in ethiopia final
Study on the human dimensions of the financial crisis in ethiopia final
 
The global financial crisis in colombia and the international conference on p...
The global financial crisis in colombia and the international conference on p...The global financial crisis in colombia and the international conference on p...
The global financial crisis in colombia and the international conference on p...
 
Monitoring the implications of the global financial crisis on primary schools...
Monitoring the implications of the global financial crisis on primary schools...Monitoring the implications of the global financial crisis on primary schools...
Monitoring the implications of the global financial crisis on primary schools...
 
A visual analytics approach to understanding poverty assessment through disas...
A visual analytics approach to understanding poverty assessment through disas...A visual analytics approach to understanding poverty assessment through disas...
A visual analytics approach to understanding poverty assessment through disas...
 
Supporting the Post-2015 Development Agenda Consultations Using U-Report - Pr...
Supporting the Post-2015 Development Agenda Consultations Using U-Report - Pr...Supporting the Post-2015 Development Agenda Consultations Using U-Report - Pr...
Supporting the Post-2015 Development Agenda Consultations Using U-Report - Pr...
 
Mobile Data for Development Primer
Mobile Data for Development PrimerMobile Data for Development Primer
Mobile Data for Development Primer
 
Big Data and the SDGs
Big Data and the SDGsBig Data and the SDGs
Big Data and the SDGs
 

Similar to GlobalPulse_SAS_MethodsPaper2011

POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOKPOLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOKIJwest
 
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
POLITICAL OPINION ANALYSIS IN SOCIAL  NETWORKS: CASE OF TWITTER AND FACEBOOK POLITICAL OPINION ANALYSIS IN SOCIAL  NETWORKS: CASE OF TWITTER AND FACEBOOK
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK dannyijwest
 
The impact of sentiment analysis from user on Facebook to enhanced the servic...
The impact of sentiment analysis from user on Facebook to enhanced the servic...The impact of sentiment analysis from user on Facebook to enhanced the servic...
The impact of sentiment analysis from user on Facebook to enhanced the servic...IJECEIAES
 
Big data analysis of news and social media content
Big data analysis of news and social media contentBig data analysis of news and social media content
Big data analysis of news and social media contentFiras Husseini
 
This is the book to use for this assignment. I am sure you probabl.docx
This is the book to use for this assignment. I am sure you probabl.docxThis is the book to use for this assignment. I am sure you probabl.docx
This is the book to use for this assignment. I am sure you probabl.docxjuliennehar
 
This is the book to use for this assignment. I am sure you probabl.docx
This is the book to use for this assignment. I am sure you probabl.docxThis is the book to use for this assignment. I am sure you probabl.docx
This is the book to use for this assignment. I am sure you probabl.docxkbrenda
 
Social implications of social networking sites
Social implications of social networking sitesSocial implications of social networking sites
Social implications of social networking sitesPetter Bae Brandtzæg
 
Social Networking Research in Europe
Social Networking Research in EuropeSocial Networking Research in Europe
Social Networking Research in EuropeDilara Adaylar
 
The Pessimistic Investor Sentiments Indicator in Social Networks
The Pessimistic Investor Sentiments Indicator in Social NetworksThe Pessimistic Investor Sentiments Indicator in Social Networks
The Pessimistic Investor Sentiments Indicator in Social NetworksTELKOMNIKA JOURNAL
 
IRJET - Social Network Stress Analysis using Word Embedding Technique
IRJET - Social Network Stress Analysis using Word Embedding TechniqueIRJET - Social Network Stress Analysis using Word Embedding Technique
IRJET - Social Network Stress Analysis using Word Embedding TechniqueIRJET Journal
 
IRJET- Sentiment Analysis using Machine Learning
IRJET- Sentiment Analysis using Machine LearningIRJET- Sentiment Analysis using Machine Learning
IRJET- Sentiment Analysis using Machine LearningIRJET Journal
 
Delaunay Triangulation Algorithms
Delaunay Triangulation AlgorithmsDelaunay Triangulation Algorithms
Delaunay Triangulation AlgorithmsAlyssa Jones
 
Saura, palos sanchez & velicia, 2020 frontiers
Saura, palos sanchez & velicia, 2020 frontiersSaura, palos sanchez & velicia, 2020 frontiers
Saura, palos sanchez & velicia, 2020 frontiersppalos68
 
EXPLORING SENTIMENT ANALYSIS RESEARCH: A SOCIAL MEDIA DATA PERSPECTIVE
EXPLORING SENTIMENT ANALYSIS RESEARCH: A SOCIAL MEDIA DATA PERSPECTIVEEXPLORING SENTIMENT ANALYSIS RESEARCH: A SOCIAL MEDIA DATA PERSPECTIVE
EXPLORING SENTIMENT ANALYSIS RESEARCH: A SOCIAL MEDIA DATA PERSPECTIVEijsc
 
A Review On Sentiment Analysis And Emotion Detection From Text
A Review On Sentiment Analysis And Emotion Detection From TextA Review On Sentiment Analysis And Emotion Detection From Text
A Review On Sentiment Analysis And Emotion Detection From TextYasmine Anino
 
Sentiment analysis of comments in social media
Sentiment analysis of comments in social media Sentiment analysis of comments in social media
Sentiment analysis of comments in social media IJECEIAES
 
A Comprehensive Study on Social Network Mental Disorders Detection
A Comprehensive Study on Social Network Mental Disorders DetectionA Comprehensive Study on Social Network Mental Disorders Detection
A Comprehensive Study on Social Network Mental Disorders Detectionijtsrd
 
System support for IM in humanitarian operations (manuscrip 18 mars)
System support for IM in humanitarian operations (manuscrip 18 mars)System support for IM in humanitarian operations (manuscrip 18 mars)
System support for IM in humanitarian operations (manuscrip 18 mars)Chalmers University of Technology
 

Similar to GlobalPulse_SAS_MethodsPaper2011 (20)

POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOKPOLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
 
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
POLITICAL OPINION ANALYSIS IN SOCIAL  NETWORKS: CASE OF TWITTER AND FACEBOOK POLITICAL OPINION ANALYSIS IN SOCIAL  NETWORKS: CASE OF TWITTER AND FACEBOOK
POLITICAL OPINION ANALYSIS IN SOCIAL NETWORKS: CASE OF TWITTER AND FACEBOOK
 
The impact of sentiment analysis from user on Facebook to enhanced the servic...
The impact of sentiment analysis from user on Facebook to enhanced the servic...The impact of sentiment analysis from user on Facebook to enhanced the servic...
The impact of sentiment analysis from user on Facebook to enhanced the servic...
 
Big data analysis of news and social media content
Big data analysis of news and social media contentBig data analysis of news and social media content
Big data analysis of news and social media content
 
H018144450
H018144450H018144450
H018144450
 
This is the book to use for this assignment. I am sure you probabl.docx
This is the book to use for this assignment. I am sure you probabl.docxThis is the book to use for this assignment. I am sure you probabl.docx
This is the book to use for this assignment. I am sure you probabl.docx
 
This is the book to use for this assignment. I am sure you probabl.docx
This is the book to use for this assignment. I am sure you probabl.docxThis is the book to use for this assignment. I am sure you probabl.docx
This is the book to use for this assignment. I am sure you probabl.docx
 
Social implications of social networking sites
Social implications of social networking sitesSocial implications of social networking sites
Social implications of social networking sites
 
Social Networking Research in Europe
Social Networking Research in EuropeSocial Networking Research in Europe
Social Networking Research in Europe
 
The Pessimistic Investor Sentiments Indicator in Social Networks
The Pessimistic Investor Sentiments Indicator in Social NetworksThe Pessimistic Investor Sentiments Indicator in Social Networks
The Pessimistic Investor Sentiments Indicator in Social Networks
 
IRJET - Social Network Stress Analysis using Word Embedding Technique
IRJET - Social Network Stress Analysis using Word Embedding TechniqueIRJET - Social Network Stress Analysis using Word Embedding Technique
IRJET - Social Network Stress Analysis using Word Embedding Technique
 
IRJET- Sentiment Analysis using Machine Learning
IRJET- Sentiment Analysis using Machine LearningIRJET- Sentiment Analysis using Machine Learning
IRJET- Sentiment Analysis using Machine Learning
 
Delaunay Triangulation Algorithms
Delaunay Triangulation AlgorithmsDelaunay Triangulation Algorithms
Delaunay Triangulation Algorithms
 
Saura, palos sanchez & velicia, 2020 frontiers
Saura, palos sanchez & velicia, 2020 frontiersSaura, palos sanchez & velicia, 2020 frontiers
Saura, palos sanchez & velicia, 2020 frontiers
 
NLP journal paper
NLP journal paperNLP journal paper
NLP journal paper
 
EXPLORING SENTIMENT ANALYSIS RESEARCH: A SOCIAL MEDIA DATA PERSPECTIVE
EXPLORING SENTIMENT ANALYSIS RESEARCH: A SOCIAL MEDIA DATA PERSPECTIVEEXPLORING SENTIMENT ANALYSIS RESEARCH: A SOCIAL MEDIA DATA PERSPECTIVE
EXPLORING SENTIMENT ANALYSIS RESEARCH: A SOCIAL MEDIA DATA PERSPECTIVE
 
A Review On Sentiment Analysis And Emotion Detection From Text
A Review On Sentiment Analysis And Emotion Detection From TextA Review On Sentiment Analysis And Emotion Detection From Text
A Review On Sentiment Analysis And Emotion Detection From Text
 
Sentiment analysis of comments in social media
Sentiment analysis of comments in social media Sentiment analysis of comments in social media
Sentiment analysis of comments in social media
 
A Comprehensive Study on Social Network Mental Disorders Detection
A Comprehensive Study on Social Network Mental Disorders DetectionA Comprehensive Study on Social Network Mental Disorders Detection
A Comprehensive Study on Social Network Mental Disorders Detection
 
System support for IM in humanitarian operations (manuscrip 18 mars)
System support for IM in humanitarian operations (manuscrip 18 mars)System support for IM in humanitarian operations (manuscrip 18 mars)
System support for IM in humanitarian operations (manuscrip 18 mars)
 

More from UN Global Pulse

Step 2: Due Diligence Questionnaire for Prospective Partners
Step 2: Due Diligence Questionnaire for Prospective PartnersStep 2: Due Diligence Questionnaire for Prospective Partners
Step 2: Due Diligence Questionnaire for Prospective PartnersUN Global Pulse
 
Step 1: Due Diligence Checklist for Prospective Partners
Step 1: Due Diligence Checklist for Prospective Partners Step 1: Due Diligence Checklist for Prospective Partners
Step 1: Due Diligence Checklist for Prospective Partners UN Global Pulse
 
Using Data and New Technology for Peacemaking, Preventive Diplomacy, and Peac...
Using Data and New Technology for Peacemaking, Preventive Diplomacy, and Peac...Using Data and New Technology for Peacemaking, Preventive Diplomacy, and Peac...
Using Data and New Technology for Peacemaking, Preventive Diplomacy, and Peac...UN Global Pulse
 
Pulse Lab Kampala Progress Report 2016/2017
Pulse Lab Kampala Progress Report 2016/2017Pulse Lab Kampala Progress Report 2016/2017
Pulse Lab Kampala Progress Report 2016/2017UN Global Pulse
 
UN Global Pulse Annual Report 2018
UN Global Pulse Annual Report 2018UN Global Pulse Annual Report 2018
UN Global Pulse Annual Report 2018UN Global Pulse
 
Pulse Lab Jakarta Annual Report 2018
Pulse Lab Jakarta Annual Report 2018 Pulse Lab Jakarta Annual Report 2018
Pulse Lab Jakarta Annual Report 2018 UN Global Pulse
 
Risks, Harms and Benefits Assessment Tool (Updated as of Jan 2019)
Risks, Harms and Benefits Assessment Tool (Updated as of Jan 2019)Risks, Harms and Benefits Assessment Tool (Updated as of Jan 2019)
Risks, Harms and Benefits Assessment Tool (Updated as of Jan 2019)UN Global Pulse
 
Pulse Lab Jakarta 2015 Annual Report
Pulse Lab Jakarta 2015 Annual Report Pulse Lab Jakarta 2015 Annual Report
Pulse Lab Jakarta 2015 Annual Report UN Global Pulse
 
Embracing Innovation: How a Social Lab can Support the Innovation Agenda in S...
Embracing Innovation: How a Social Lab can Support the Innovation Agenda in S...Embracing Innovation: How a Social Lab can Support the Innovation Agenda in S...
Embracing Innovation: How a Social Lab can Support the Innovation Agenda in S...UN Global Pulse
 
Urban Vulnerability Mapping Toolkit
Urban Vulnerability Mapping ToolkitUrban Vulnerability Mapping Toolkit
Urban Vulnerability Mapping ToolkitUN Global Pulse
 
Navigating the Terrain: A Toolkit for Conceptualising Service Design Projects
Navigating the Terrain: A Toolkit for Conceptualising Service Design ProjectsNavigating the Terrain: A Toolkit for Conceptualising Service Design Projects
Navigating the Terrain: A Toolkit for Conceptualising Service Design ProjectsUN Global Pulse
 
Experimenting with Big Data and AI to Support Peace and Security
Experimenting with Big Data and AI to Support Peace and SecurityExperimenting with Big Data and AI to Support Peace and Security
Experimenting with Big Data and AI to Support Peace and SecurityUN Global Pulse
 
Banking on Fintech: Financial inclusion for micro enterprises in Indonesia
Banking on Fintech: Financial inclusion for micro enterprises in IndonesiaBanking on Fintech: Financial inclusion for micro enterprises in Indonesia
Banking on Fintech: Financial inclusion for micro enterprises in IndonesiaUN Global Pulse
 
Haze Gazer - Tool Overview
Haze Gazer - Tool Overview Haze Gazer - Tool Overview
Haze Gazer - Tool Overview UN Global Pulse
 
Building Proxy Indicators of National Wellbeing with Postal Data - Project Ov...
Building Proxy Indicators of National Wellbeing with Postal Data - Project Ov...Building Proxy Indicators of National Wellbeing with Postal Data - Project Ov...
Building Proxy Indicators of National Wellbeing with Postal Data - Project Ov...UN Global Pulse
 
Sex Disaggregation of Social Media Posts - Tool Overview
Sex Disaggregation of Social Media Posts - Tool OverviewSex Disaggregation of Social Media Posts - Tool Overview
Sex Disaggregation of Social Media Posts - Tool OverviewUN Global Pulse
 
Using Big data Analytics for Improved Public Transport
Using Big data Analytics for Improved Public Transport Using Big data Analytics for Improved Public Transport
Using Big data Analytics for Improved Public Transport UN Global Pulse
 
Translator Gator - Tool Overview
Translator Gator - Tool Overview Translator Gator - Tool Overview
Translator Gator - Tool Overview UN Global Pulse
 
Big Data for Financial Inclusion, Examining the Customer Journey - Project Ov...
Big Data for Financial Inclusion, Examining the Customer Journey - Project Ov...Big Data for Financial Inclusion, Examining the Customer Journey - Project Ov...
Big Data for Financial Inclusion, Examining the Customer Journey - Project Ov...UN Global Pulse
 
Understanding Perceptions of Migrants and Refugees with Social Media - Projec...
Understanding Perceptions of Migrants and Refugees with Social Media - Projec...Understanding Perceptions of Migrants and Refugees with Social Media - Projec...
Understanding Perceptions of Migrants and Refugees with Social Media - Projec...UN Global Pulse
 

More from UN Global Pulse (20)

Step 2: Due Diligence Questionnaire for Prospective Partners
Step 2: Due Diligence Questionnaire for Prospective PartnersStep 2: Due Diligence Questionnaire for Prospective Partners
Step 2: Due Diligence Questionnaire for Prospective Partners
 
Step 1: Due Diligence Checklist for Prospective Partners
Step 1: Due Diligence Checklist for Prospective Partners Step 1: Due Diligence Checklist for Prospective Partners
Step 1: Due Diligence Checklist for Prospective Partners
 
Using Data and New Technology for Peacemaking, Preventive Diplomacy, and Peac...
Using Data and New Technology for Peacemaking, Preventive Diplomacy, and Peac...Using Data and New Technology for Peacemaking, Preventive Diplomacy, and Peac...
Using Data and New Technology for Peacemaking, Preventive Diplomacy, and Peac...
 
Pulse Lab Kampala Progress Report 2016/2017
Pulse Lab Kampala Progress Report 2016/2017Pulse Lab Kampala Progress Report 2016/2017
Pulse Lab Kampala Progress Report 2016/2017
 
UN Global Pulse Annual Report 2018
UN Global Pulse Annual Report 2018UN Global Pulse Annual Report 2018
UN Global Pulse Annual Report 2018
 
Pulse Lab Jakarta Annual Report 2018
Pulse Lab Jakarta Annual Report 2018 Pulse Lab Jakarta Annual Report 2018
Pulse Lab Jakarta Annual Report 2018
 
Risks, Harms and Benefits Assessment Tool (Updated as of Jan 2019)
Risks, Harms and Benefits Assessment Tool (Updated as of Jan 2019)Risks, Harms and Benefits Assessment Tool (Updated as of Jan 2019)
Risks, Harms and Benefits Assessment Tool (Updated as of Jan 2019)
 
Pulse Lab Jakarta 2015 Annual Report
Pulse Lab Jakarta 2015 Annual Report Pulse Lab Jakarta 2015 Annual Report
Pulse Lab Jakarta 2015 Annual Report
 
Embracing Innovation: How a Social Lab can Support the Innovation Agenda in S...
Embracing Innovation: How a Social Lab can Support the Innovation Agenda in S...Embracing Innovation: How a Social Lab can Support the Innovation Agenda in S...
Embracing Innovation: How a Social Lab can Support the Innovation Agenda in S...
 
Urban Vulnerability Mapping Toolkit
Urban Vulnerability Mapping ToolkitUrban Vulnerability Mapping Toolkit
Urban Vulnerability Mapping Toolkit
 
Navigating the Terrain: A Toolkit for Conceptualising Service Design Projects
Navigating the Terrain: A Toolkit for Conceptualising Service Design ProjectsNavigating the Terrain: A Toolkit for Conceptualising Service Design Projects
Navigating the Terrain: A Toolkit for Conceptualising Service Design Projects
 
Experimenting with Big Data and AI to Support Peace and Security
Experimenting with Big Data and AI to Support Peace and SecurityExperimenting with Big Data and AI to Support Peace and Security
Experimenting with Big Data and AI to Support Peace and Security
 
Banking on Fintech: Financial inclusion for micro enterprises in Indonesia
Banking on Fintech: Financial inclusion for micro enterprises in IndonesiaBanking on Fintech: Financial inclusion for micro enterprises in Indonesia
Banking on Fintech: Financial inclusion for micro enterprises in Indonesia
 
Haze Gazer - Tool Overview
Haze Gazer - Tool Overview Haze Gazer - Tool Overview
Haze Gazer - Tool Overview
 
Building Proxy Indicators of National Wellbeing with Postal Data - Project Ov...
Building Proxy Indicators of National Wellbeing with Postal Data - Project Ov...Building Proxy Indicators of National Wellbeing with Postal Data - Project Ov...
Building Proxy Indicators of National Wellbeing with Postal Data - Project Ov...
 
Sex Disaggregation of Social Media Posts - Tool Overview
Sex Disaggregation of Social Media Posts - Tool OverviewSex Disaggregation of Social Media Posts - Tool Overview
Sex Disaggregation of Social Media Posts - Tool Overview
 
Using Big data Analytics for Improved Public Transport
Using Big data Analytics for Improved Public Transport Using Big data Analytics for Improved Public Transport
Using Big data Analytics for Improved Public Transport
 
Translator Gator - Tool Overview
Translator Gator - Tool Overview Translator Gator - Tool Overview
Translator Gator - Tool Overview
 
Big Data for Financial Inclusion, Examining the Customer Journey - Project Ov...
Big Data for Financial Inclusion, Examining the Customer Journey - Project Ov...Big Data for Financial Inclusion, Examining the Customer Journey - Project Ov...
Big Data for Financial Inclusion, Examining the Customer Journey - Project Ov...
 
Understanding Perceptions of Migrants and Refugees with Social Media - Projec...
Understanding Perceptions of Migrants and Refugees with Social Media - Projec...Understanding Perceptions of Migrants and Refugees with Social Media - Projec...
Understanding Perceptions of Migrants and Refugees with Social Media - Projec...
 

Recently uploaded

The New Cloud World Order Is FinOps (Slideshow)
The New Cloud World Order Is FinOps (Slideshow)The New Cloud World Order Is FinOps (Slideshow)
The New Cloud World Order Is FinOps (Slideshow)codyslingerland1
 
Technical SEO for Improved Accessibility WTS FEST
Technical SEO for Improved Accessibility  WTS FESTTechnical SEO for Improved Accessibility  WTS FEST
Technical SEO for Improved Accessibility WTS FESTBillieHyde
 
Introduction - IPLOOK NETWORKS CO., LTD.
Introduction - IPLOOK NETWORKS CO., LTD.Introduction - IPLOOK NETWORKS CO., LTD.
Introduction - IPLOOK NETWORKS CO., LTD.IPLOOK Networks
 
SIM INFORMATION SYSTEM: REVOLUTIONIZING DATA MANAGEMENT
SIM INFORMATION SYSTEM: REVOLUTIONIZING DATA MANAGEMENTSIM INFORMATION SYSTEM: REVOLUTIONIZING DATA MANAGEMENT
SIM INFORMATION SYSTEM: REVOLUTIONIZING DATA MANAGEMENTxtailishbaloch
 
IT Service Management (ITSM) Best Practices for Advanced Computing
IT Service Management (ITSM) Best Practices for Advanced ComputingIT Service Management (ITSM) Best Practices for Advanced Computing
IT Service Management (ITSM) Best Practices for Advanced ComputingMAGNIntelligence
 
Automation Ops Series: Session 2 - Governance for UiPath projects
Automation Ops Series: Session 2 - Governance for UiPath projectsAutomation Ops Series: Session 2 - Governance for UiPath projects
Automation Ops Series: Session 2 - Governance for UiPath projectsDianaGray10
 
Extra-120324-Visite-Entreprise-icare.pdf
Extra-120324-Visite-Entreprise-icare.pdfExtra-120324-Visite-Entreprise-icare.pdf
Extra-120324-Visite-Entreprise-icare.pdfInfopole1
 
Top 10 Squarespace Development Companies
Top 10 Squarespace Development CompaniesTop 10 Squarespace Development Companies
Top 10 Squarespace Development CompaniesTopCSSGallery
 
TrustArc Webinar - How to Live in a Post Third-Party Cookie World
TrustArc Webinar - How to Live in a Post Third-Party Cookie WorldTrustArc Webinar - How to Live in a Post Third-Party Cookie World
TrustArc Webinar - How to Live in a Post Third-Party Cookie WorldTrustArc
 
My key hands-on projects in Quantum, and QAI
My key hands-on projects in Quantum, and QAIMy key hands-on projects in Quantum, and QAI
My key hands-on projects in Quantum, and QAIVijayananda Mohire
 
Design and Modeling for MySQL SCALE 21X Pasadena, CA Mar 2024
Design and Modeling for MySQL SCALE 21X Pasadena, CA Mar 2024Design and Modeling for MySQL SCALE 21X Pasadena, CA Mar 2024
Design and Modeling for MySQL SCALE 21X Pasadena, CA Mar 2024Alkin Tezuysal
 
EMEA What is ThousandEyes? Webinar
EMEA What is ThousandEyes? WebinarEMEA What is ThousandEyes? Webinar
EMEA What is ThousandEyes? WebinarThousandEyes
 
AI Workshops at Computers In Libraries 2024
AI Workshops at Computers In Libraries 2024AI Workshops at Computers In Libraries 2024
AI Workshops at Computers In Libraries 2024Brian Pichman
 
From the origin to the future of Open Source model and business
From the origin to the future of  Open Source model and businessFrom the origin to the future of  Open Source model and business
From the origin to the future of Open Source model and businessFrancesco Corti
 
Graphene Quantum Dots-Based Composites for Biomedical Applications
Graphene Quantum Dots-Based Composites for  Biomedical ApplicationsGraphene Quantum Dots-Based Composites for  Biomedical Applications
Graphene Quantum Dots-Based Composites for Biomedical Applicationsnooralam814309
 
GraphSummit Copenhagen 2024 - Neo4j Vision and Roadmap.pptx
GraphSummit Copenhagen 2024 - Neo4j Vision and Roadmap.pptxGraphSummit Copenhagen 2024 - Neo4j Vision and Roadmap.pptx
GraphSummit Copenhagen 2024 - Neo4j Vision and Roadmap.pptxNeo4j
 
Stobox 4: Revolutionizing Investment in Real-World Assets Through Tokenization
Stobox 4: Revolutionizing Investment in Real-World Assets Through TokenizationStobox 4: Revolutionizing Investment in Real-World Assets Through Tokenization
Stobox 4: Revolutionizing Investment in Real-World Assets Through TokenizationStobox
 
UiPath Studio Web workshop series - Day 4
UiPath Studio Web workshop series - Day 4UiPath Studio Web workshop series - Day 4
UiPath Studio Web workshop series - Day 4DianaGray10
 
Emil Eifrem at GraphSummit Copenhagen 2024 - The Art of the Possible.pptx
Emil Eifrem at GraphSummit Copenhagen 2024 - The Art of the Possible.pptxEmil Eifrem at GraphSummit Copenhagen 2024 - The Art of the Possible.pptx
Emil Eifrem at GraphSummit Copenhagen 2024 - The Art of the Possible.pptxNeo4j
 
Q4 2023 Quarterly Investor Presentation - FINAL - v1.pdf
Q4 2023 Quarterly Investor Presentation - FINAL - v1.pdfQ4 2023 Quarterly Investor Presentation - FINAL - v1.pdf
Q4 2023 Quarterly Investor Presentation - FINAL - v1.pdfTejal81
 

Recently uploaded (20)

The New Cloud World Order Is FinOps (Slideshow)
The New Cloud World Order Is FinOps (Slideshow)The New Cloud World Order Is FinOps (Slideshow)
The New Cloud World Order Is FinOps (Slideshow)
 
Technical SEO for Improved Accessibility WTS FEST
Technical SEO for Improved Accessibility  WTS FESTTechnical SEO for Improved Accessibility  WTS FEST
Technical SEO for Improved Accessibility WTS FEST
 
Introduction - IPLOOK NETWORKS CO., LTD.
Introduction - IPLOOK NETWORKS CO., LTD.Introduction - IPLOOK NETWORKS CO., LTD.
Introduction - IPLOOK NETWORKS CO., LTD.
 
SIM INFORMATION SYSTEM: REVOLUTIONIZING DATA MANAGEMENT
SIM INFORMATION SYSTEM: REVOLUTIONIZING DATA MANAGEMENTSIM INFORMATION SYSTEM: REVOLUTIONIZING DATA MANAGEMENT
SIM INFORMATION SYSTEM: REVOLUTIONIZING DATA MANAGEMENT
 
IT Service Management (ITSM) Best Practices for Advanced Computing
IT Service Management (ITSM) Best Practices for Advanced ComputingIT Service Management (ITSM) Best Practices for Advanced Computing
IT Service Management (ITSM) Best Practices for Advanced Computing
 
Automation Ops Series: Session 2 - Governance for UiPath projects
Automation Ops Series: Session 2 - Governance for UiPath projectsAutomation Ops Series: Session 2 - Governance for UiPath projects
Automation Ops Series: Session 2 - Governance for UiPath projects
 
Extra-120324-Visite-Entreprise-icare.pdf
Extra-120324-Visite-Entreprise-icare.pdfExtra-120324-Visite-Entreprise-icare.pdf
Extra-120324-Visite-Entreprise-icare.pdf
 
Top 10 Squarespace Development Companies
Top 10 Squarespace Development CompaniesTop 10 Squarespace Development Companies
Top 10 Squarespace Development Companies
 
TrustArc Webinar - How to Live in a Post Third-Party Cookie World
TrustArc Webinar - How to Live in a Post Third-Party Cookie WorldTrustArc Webinar - How to Live in a Post Third-Party Cookie World
TrustArc Webinar - How to Live in a Post Third-Party Cookie World
 
My key hands-on projects in Quantum, and QAI
My key hands-on projects in Quantum, and QAIMy key hands-on projects in Quantum, and QAI
My key hands-on projects in Quantum, and QAI
 
Design and Modeling for MySQL SCALE 21X Pasadena, CA Mar 2024
Design and Modeling for MySQL SCALE 21X Pasadena, CA Mar 2024Design and Modeling for MySQL SCALE 21X Pasadena, CA Mar 2024
Design and Modeling for MySQL SCALE 21X Pasadena, CA Mar 2024
 
EMEA What is ThousandEyes? Webinar
EMEA What is ThousandEyes? WebinarEMEA What is ThousandEyes? Webinar
EMEA What is ThousandEyes? Webinar
 
AI Workshops at Computers In Libraries 2024
AI Workshops at Computers In Libraries 2024AI Workshops at Computers In Libraries 2024
AI Workshops at Computers In Libraries 2024
 
From the origin to the future of Open Source model and business
From the origin to the future of  Open Source model and businessFrom the origin to the future of  Open Source model and business
From the origin to the future of Open Source model and business
 
Graphene Quantum Dots-Based Composites for Biomedical Applications
Graphene Quantum Dots-Based Composites for  Biomedical ApplicationsGraphene Quantum Dots-Based Composites for  Biomedical Applications
Graphene Quantum Dots-Based Composites for Biomedical Applications
 
GraphSummit Copenhagen 2024 - Neo4j Vision and Roadmap.pptx
GraphSummit Copenhagen 2024 - Neo4j Vision and Roadmap.pptxGraphSummit Copenhagen 2024 - Neo4j Vision and Roadmap.pptx
GraphSummit Copenhagen 2024 - Neo4j Vision and Roadmap.pptx
 
Stobox 4: Revolutionizing Investment in Real-World Assets Through Tokenization
Stobox 4: Revolutionizing Investment in Real-World Assets Through TokenizationStobox 4: Revolutionizing Investment in Real-World Assets Through Tokenization
Stobox 4: Revolutionizing Investment in Real-World Assets Through Tokenization
 
UiPath Studio Web workshop series - Day 4
UiPath Studio Web workshop series - Day 4UiPath Studio Web workshop series - Day 4
UiPath Studio Web workshop series - Day 4
 
Emil Eifrem at GraphSummit Copenhagen 2024 - The Art of the Possible.pptx
Emil Eifrem at GraphSummit Copenhagen 2024 - The Art of the Possible.pptxEmil Eifrem at GraphSummit Copenhagen 2024 - The Art of the Possible.pptx
Emil Eifrem at GraphSummit Copenhagen 2024 - The Art of the Possible.pptx
 
Q4 2023 Quarterly Investor Presentation - FINAL - v1.pdf
Q4 2023 Quarterly Investor Presentation - FINAL - v1.pdfQ4 2023 Quarterly Investor Presentation - FINAL - v1.pdf
Q4 2023 Quarterly Investor Presentation - FINAL - v1.pdf
 

GlobalPulse_SAS_MethodsPaper2011

  • 1.       USING  SOCIAL  MEDIA  AND  ONLINE  CONVERSATIONS   TO  ADD  DEPTH  TO  UNEMPLOYMENT  STATISTICS   Methodological  White  Paper1   December  8,  2011     Abstract   This   research   investigates   whether   and   how   social   media   and   other   online   user-­‐generated   content   could   enrich   understanding   of   the   effect   of   changing   employment   conditions.   The   primary   goal   is   to   compare   the   qualitative   information   offered   by   social   media   with   unemployment   figures.   To   this   end,   we   first   selected   online   job-­‐related   conversations   from   blogs,   forums   and   news   from   the   United   States   and   Ireland.   For   all   documents,   a   quantitative   mood   score   based   on   the   tone   of   the   conversations—for   example   happiness,   depression   or   anxiety—  was  assigned.  The  number  of  unemployment-­‐related  documents  that  also  dealt  with   other  topics  such  as  housing  and  transportation  was  also  quantified,  in  order  to  gain  insight  into     populations’   coping   mechanisms.   This   data   was   analyzed   in   two   primary   ways.   First,   the   quantified   mood   scoring   was   correlated   to   the   unemployment   rate   to   discover   leading   indicators   that   forecast   rises   and   falls   in   the   unemployment   rate.   For   example,   the   volume   of   conversations   in   Ireland   categorized   as   showing   a   confused   mood   correlated   with   the   unemployment   rate   with   a   lead-­‐time   of   three   (3)   months.   Second,   the   volume   of   documents   related   to   coping   mechanisms   also   showed   a   significant   relationship   with   the   unemployment   rate,  which  may  give  insight  into  the  reactions  that  can  be  expected  from  a  population  dealing   with   unemployment.   For   example,   the   conversations   in   the   US   around   the   loss   of   housing   increased   two   (2)   months   after   unemployment   spikes.   Overall,   in   this   initial   research,  SAS   and   Global   Pulse   have   underlined   the   potential   of   online   conversations   to   complement   official   statistics,   by   providing   a   qualitative   picture   demonstrating   how   people   are   feeling   and   coping   with  respect  to  their  employment  status.   1. Background  and  Objective   Global   Pulse   is   dedicated   to   better   understanding   how   new   types   of   data   can   strengthen   available   information   on   how   people   are   impacted   by   global   crises.   This   project,   conducted   in   partnership   with   SAS,   seeks   to   lay   a   foundation   to   use   what   Global   Pulse   believes   could   represent  a  powerful  source  of  new  data:  the  global  conversation  that  is  taking  place  over  social   media  and  online  content.                                                                                                                             1  This methods white paper arose from an on-going series of collaborative research projects conducted by the United Nations Global Pulse in 2011. Global Pulse is an innovation initiative of the Executive Office of the UN Secretary-General, which seeks to harness the opportunities in digital data to strengthen evidence- based decision-making. This research was designed to better understand where digital data can add value to existing policy analysis, and to contribute to future applications of digital data to global development. This project was conducted in collaboration with SAS. For more information on this project or the other projects in this series, please visit: http://www.unglobalpulse.org/research.   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   1      
  • 2.       In  particular,  this  research  focuses  on  what  unemployment  looks  like  in  social  media,  specifically   from  June  2009  to  June  2011.     This  project  investigates  whether  and  how  the  feelings  and  information  shared  on  social  media   and  other,  online,  user-­‐generated  content  could  enrich  understanding  of  the  effect  of  changing   employment  conditions  on  people’s  perceptions  and  decisions.     The  research  zoomed  in  on  the  cases  of  the  US  and  Ireland,  where  social  media  and  blogs  are   widely   used   and   where   the   recent   and   ongoing   economic   crisis   has   had   a   severe   impact   on   employment.   In   particular,   the   robust   economy   and   low   unemployment   in   both   the   US   and   Ireland  has,  over  the  past  two  to  three  years,  given  way  to  unprecedented  high  unemployment   rates  (see  Figure  1).  In  Ireland,  according  to  EuroStat,  the  unemployment  rate  has  been  under   5%   since   2000;   since   the   economic   crisis,   the   rate   has   risen   to   as   high   as   15%   and   fluctuated   around   14%,   a   level   not   seen   since   the   early   90s.   According   to   the   US   Bureau   of   Labor   Statistics,   the  unemployment  rate  in  the  US  reached  double  digits  for  the  first  time  since  the  early  80s,  and   has   dropped   slightly   from   that   peak   to   hover   around   9%.   The   organization   responsible   for   providing  the  Ireland  unemployment  rates  to  the  database  is  the  Central  Statistics  Office,  Dublin   and  for  the  United  States,  the  data  was  provided  by  the  United  States  Census  Bureau.     In   this   exploratory   project,   two   specific   questions   were   addressed:   can   online   conversations   provide   an   early   indicator   of   impending   job   losses,   and   can   they   help   policy   makers   enrich   their  understanding  of  the  type  and  sequence  of  coping  strategies  employed  by  individuals?     In  order  to  answer  these  questions,  job-­‐related  user-­‐generated  unstructured  text  data  was   analyzed  using  SAS  tools  and  technologies  (see  Annex  A  for  a  detailed  list).  This  allowed   researchers  to  first  retrieve  pertinent  documents;  second,  analyze  those  documents  to  quantify   the  feelings,  moods,  concerns  and  strategies;  and  finally  to  correlate  the  data  with  official   unemployment  statistics.     White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   2      
  • 3.       a Public  Data: Unemployment-­‐ Blogs,  Forums,  News related   Conversations US  /  Ireland b Sentiments Topics c Official   Unemployment   Statistics   Figure  1:  Project  Workflow   2. Process  Overview   The  project  was  implemented  according  to  following  workflow  (see  Figure  1):   a) Online   job-­‐related   conversations   from   blogs,   forums   and   news   were   automatically   retrieved.   b) Each  document  was  assigned  a  quantitative  mood  score  based  on  the  tone  or  mood  of   the   conversations—for   example   happiness,   depression,   anxiety—it   contained.   The   number  of  unemployment  related  documents  that  also  dealt  with  other  topics,  such  as   housing   and   transportation,   was   quantified   and   categorized   into   pre-­‐defined   lists   of   document  topics  representing  potential  coping  mechanisms.   c) These   measures—aggregated   mood   scores   and   the   volume   of   conversations   around   different   topics—   were   compared   with   official   unemployment   statistics   over   time   in   search  of  interesting  dynamic-­‐correlations.   3. Data  Acquisition  and  Treatment   3.1. Acquisition   This  process  defines  the  relevant  data  set.  The  sample  data  was  pulled  from  a  variety  of  public   Internet  sources,  i.e.  blogs,  internet  forums  and  news,  published  in  the  US  and  Ireland  over  the   past   two   years.   Relevant   documents   containing   references   to   terms   defining   unemployment,   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   3      
  • 4.       specifically   terms   and   phrases   such   as   “unemployed,   fired,   on   the   dole,   and   collecting   unemployment”,   were   automatically   retrieved   and   identified.   The   selection   was   subsequently   refined  using  additional  unemployment  synonyms.   3.2. Filtering   Once  the  data  set  is  defined,  the  documents  pass  through  a  custom  filtering  process  meant  to   eliminate  noise  and  sort  the  documents  by  topic.   The  words  used  in  each  document  were  mined  in  order  to  assign  one  or  more  topical  categories.   These   categories   were   chosen   to   represent   some   typical   coping   mechanisms   and   included:   Housing,   Transportation,   Entertainment,   Consumer   Spending,   Bills,   Financial   Distress,   Underemployment,   Nutrition   Quality,   Alcohol,   Borrow/Save   Money,   Education,   Healthcare,   Unemployment  Claims,  and  Travel  (see  Annex  B  for  textual  examples  of  these  categories).   Categories   related   specifically   to   job   status   were   also   constructed,   such   as   the   chatter   of   people   who   downgraded   by   considering   employment   that   paid   less   or   required   lower   qualifications   than   previous   employment.   Finally,   underemployment,   where   people   are   working   only   part   time,  was  also  examined.   3.3. Validation   Selected   documents   underwent   a   series   of   validation   rounds   to   ensure   the   quality   and   accuracy   of  the  query  and  filtering  processes.  For  this  purpose,  a  representative  sample  of  the  data  was   manually   reviewed   to   check   the   relevance   of   the   content   and   the   adequacy   of   the   categorization  into  different  topical  categories.  Once  a  reasonable  number  of  documents  were   successfully   reviewed,   the   query   and   filtering   processes   were   approved   and   the   resulting   dataset  was  deemed  appropriate  for  analysis.   4. Analysis   4.1. Sentiment  analysis   Each   selected   document   was   then   evaluated   with   sentiment   analysis   techniques—a   form   of   automated   text   analysis   that   assesses   the   nature   (e.g.   anxiety,   confusion,   hostility,   etc.)   and   polarity   (positive,   negative,   neutral)   of   the   content   expressed   using   natural   language   processing   and  linguistic  rules.  Each  document  was  subsequently  assigned  a  mood  score,   according   to  six   different   sentiments   using   the   SAS’   sentiment   analysis   engine   (see   Annex   B2   for   textual   examples   of   these   categories).   When   a   word   or   phrase   within   a   document   is   a   predefined   classifier   for   one   of   the   mood   categories,   that   document   received   a   numeric   score   for   that   mood.   The  application  displays  the  various  moods  that  dominated  social  media  in  any  given  month  (as   depicted   on   the   left)   or   the   change   in   a   mood   state   over   time   (confidence   in   the   US   depicted   on   the  right).   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   4      
  • 5.                             Figure  2:  Sentiment  Analysis   This   analysis   gives   the   user   an   idea   of   how   the   unemployed   feel   about   their   status.   Are   they   optimistic   about   the   future?   Are   they   depressed   or   anxious   about   their   job   prospects?   In   this   manner,   sentiment   analysis   can   provide   an   important   complement   to   the   unemployment   statistics  by  essentially  quantifying  the  qualitative  experience  of  the  unemployed.   4.2. Cross-­‐Correlation  Analysis   Finally,  dynamic-­‐correlation  tests  between  mood  scores  and  the  volume  of  documents  in  a  given   category   with   official   unemployment   rate   were   performed   in   order   to   reveal,   in   social   media   and  online  user-­‐generated  content:   • Leading  indicators  that  could  give  advance  warning  about  unemployment  statistics   • Lagging   indicators   of   impact   and   coping   strategies   that   could   be   used   to   design   policy   interventions   aimed   at   mitigating   the   effects   of   unemployment,   i.e.   to   assess   how   various   types   of   unemployment   chatter   might   predict   the   impact   of   changing   employment   conditions   on   other   public   services,   such   as   the   use   of   public   transportation,  housing,  etc.   These   dynamic-­‐correlation   tests   were   conducted   using   the   maximum   value   of   the   average   weekly   moods   scores   assigned   to   the   corpus   of   selected   documents.   Coping   strategy   volumes   are  obtained  per   month.  These  scores  and  volumes  are  tested  for  dynamic-­‐correlation  against   the  official  unemployment  rate.  Only  results  at  a  90%  confidence  level  or  higher  are  discussed   here  and  presented  in  the  dashboard.   The   category   of   transportation   provides   one   compelling   example.   The   dynamic-­‐correlation   analysis  plot  below  shows  a  statistically  significant  relationship  between  transportation  chatter   and  the  actual  unemployment  rate.     White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   5      
  • 6.         Figure  3:  Cross  Series  Plot   Upon   further   investigation,   it   became   clear   that   discussions   about   public   transportation   spike   one  month  prior  to  a  spike  in  unemployment.  This  suggests  that  an  increased  demand  for  public   transportation  can  be  expected  to  spike  prior  to  a  spike  in  unemployment.     Figure  4:  Cross-­‐Correlations   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   6      
  • 7.       Correlation  analysis  was  conducted  for  all  of  the  mood  states  and  coping  strategies  and  resulted   in  statistically  significant  correlations  between  several  categories  and  the  official  unemployment   rate  (see  Annex  C  for  more  examples  of  cross-­‐correlation  functions).   5. Outcomes   5.1. Unemployment  Dashboard   The   SAS   team   created   two   Dashboards—one   for   Ireland   and   one   for   the   US—   that   provide   a   visual   representation   of   findings.   The   dashboards   were   designed   to   be   easily   understood   by   non-­‐experts,  and  it  is  an  example  of  how  one  might  monitor  the  real  time  breakdown  of  social   listening  results  in  the  future.   The   Dashboard   allows   the   user   to   investigate   the   volume   of   conversations   around   unemployment   as   well   as   the   coping   mechanisms   that   are   being   discussed   in   relation   to   unemployment.   The   Dashboard   also   identifies   time   relationships   between   unemployment,   conversation   moods,   coping   mechanisms   and   various   macroeconomic   indicators,   which   allow   the  user  to  see  patterns  and  predictions.  Understanding  the  timing  and  circumstances  that  are   causing  a  shift  in  a  populations’  behavior  allows  decision  makers  to  prepare  and  act.     Figure  5:  United  States  Dashboard   In  the  top  left  quadrant   The   official   unemployment   rate   and   the   forecasting   model   are   depicted.   The   forecast   of   the   unemployment   rate   was   obtained   using   the   SAS®   Time   Series   Forecasting   System.   The   model   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   7      
  • 8.       with   the   best   Akaike   information   criterion   to   measure   the   relative   goodness   of   fit   of   a   statistical   model  is  chosen.  A  Linear  Holt  Exponential  Smoothing  model  was  chosen  for  the  United  States   and  a  Winters  Method  Additive  Model  was  chosen  for  Ireland.  In  the  future,  the  results  such  as   those   from   this   proof   of   value   study   could   become   valuable   inputs   for   a   predictive   unemployment  model.  This  quadrant  exemplifies  the  possible  display  of  such  a  model.   In  the  top  right  quadrant   The  volume  of  documents  that  discuss  particular  topics  are  presented  per  month  in  a  pie  chart.   In  the  bottom  right  quadrant   The   mood   scores   per   month   are   shown   as   six   mood   dials   or   barometers.   The   benefit   of   presenting  the  mood  in  this  manner  is  that  it  allows  a  quick  determination  of  changes  over  the   past  month—in  other  words;  it  is  easy  to  see  if  there  has  been  a  negative  or  positive  change  in   mood  over  the  past  month.  This  information  is  also  available  over  time.  By  clicking  on  the  title   above   the   mood   dials,   a   new   dashboard   will   load   which   displays   the   results   for   a   selected   mood   in  a  bar  chart  format  over  time.   In  the  bottom  left  quadrant   Dynamic-­‐Correlation  results  are  shown  in  a  vertical  bar  chart.  The  time  lag  of  the  correlation  is   on  the  x-­‐axis  and  the  variable  correlated  with  the  unemployment  rate  is  on  the  y-­‐axis.   Other   features   of   the   dashboard   allow   the   user   to   understand   the   analysis   better.   Additional   links  contained  in  the  dashboard  will:   • provide  examples  of  phrases  from  actual  document  which  describe  the  coping  strategy   or  mood  state   • allow  the  user  to  explore  the  dynamic-­‐correlations  in  more  depth   These  features  allow  the  user  to  understand  better  the  nature  of  the  documents  and  the  data   that  informs  the  analysis.   Finally,  the  dashboard  allows  users  to  look  at  changes  over  time.  Selecting  a  month  changes  all   other   appropriate   visuals,   allowing   for   the   exploration   of   the   interplay   between   the   unemployment  rate,  the  mood  of  the  unemployed  and  coping  mechanisms.  The  data  and  results   are  represented  in  a  SAS®  Business  Intelligence  Dashboard  Interface.   5.2. Results   Starting   from   June   2009   to   June   2011,   the   dashboard   retrieved   over   28,000   documents   for   Ireland  and  430,000  for  the  United  States.  Over  half  of  the  documents  pulled  in  each  contained   coping  mechanism  classifiers.   The   following   figures   provide   a   graphical  example   of   the   leading   and   lagging   relationships   found   in  the  data:   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   8      
  • 9.         Figure  6:  United  States  Unemployment  Spike     Figure  7:  Ireland  Unemployment  Spike   Overall,   among   the   40+   cross-­‐correlations   explored,   five   (5)   indicators   in   the   US   and   six   (6)   indicators  in  Ireland  showed  a  correlation  significant  at  the  90%  confidence  level  or  higher.     White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   9      
  • 10.       These  findings  are  a  first  step  in  this  kind  of  research  and  represent  a  high  potential  source  of   complementary   information   to   the   unemployment   statistics.   The   following   images   depict   the   strongest  dynamic-­‐correlations  that  were  found:       Figure  8:  United  States  Chatter     Figure  9:  Ireland  Chatter   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   10      
  • 11.         Table  1:  Country  Correlation,  CCF  and  Significance  Level   Country   Correlation  Description   CCF   Significance   Level   US   Hostile  Mood  increases  4  months  before  a   .442   95%   spike  in  unemployment   US   Depressed  Mood  increases  4  months   .489   95%   before  a  spike  in  unemployment   US   Uncertain  Mood  increases  as   .448   95%   unemployment  spikes   US   Talk  about  loss  of  housing  increases  2   .634   95%   months  after  an  unemployment  spike   US   Talk  about  auto  repossession  increases  3   .455   95%   months  after  an  unemployment  spike   IRELAND   Anxious  Mood  increases  5  months  before   .387   90%   a  spike  in  unemployment   IRELAND   Confused  Mood  increases  3  months   .675   95%   before  a  spike  in  unemployment   IRELAND   Confident  Mood  decreases  2  months   -­‐.407   90%   before  a  spike  in  unemployment   IRELAND   Talk  about  changing  transportation   .380   90%   methods  for  the  worse  increases  as   unemployment  increases   IRELAND   Talk  about  travel  cancelations  increases  3   .450   95%   months  after  an  unemployment  spike   IRELAND   Talk  about  changing  housing  situations  for   .328   90%   the  worse  increases  8  months  after   unemployment  increases   6. Conclusions  and  perspectives   In   this   initial   research,   the   high   potential   of   online   conversations   to   complement   official   statistics  has  been  shown,  by  providing  leading  and  lagging  indicators  that  show  how  people  are   feeling   with   respect   to   their   employment   status   and   which   coping   strategies   (conversation   topics)  are  employed  over  time.   Several  mood  score  volumes  produced  strong  correlations  with  the  unemployment  rate,  which   may   be   leading   indicators   that   the   unemployment   will   rise   or   fall.   For   example,   conversations   in   Ireland   showing   a   confused   mood   preceded   the   unemployment   rate   variations   by   3   months.   And   data   pulled   from   the   social   listening   sources,   representing   daily   conversations,   proved   to   contain  valuable  information  related  to  how  the  unemployed  cope.  In  addition,  changes  in  the   volume   of   particular   coping   topics   showed   significant   lagging   relationships   with   the   unemployment  rate  that  may  give  insight  into  reactions  that  can  be  expected  from  a  population   dealing  with  unemployment  –  as  demonstrated,  for  example,  by  the  increase  in  conversations  in   the  US  around  the  loss  of  housing  2  months  after  an  increase  in  unemployment.     White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   11      
  • 12.       In  short,  the  type  of  data  needed  to  perform  this  sort  of  analysis  exists,  and  the  initial  research   pointed   to   potential   indicators   that   might   be   used   for   improving   unemployment   policies   or   social  protection  programs.   Several   challenges   in   this   type   of   analysis   and   future   lines   of   research   were   also   made   clear   through   the   course   of   this   work.   First,   greater   geographical   information   related   to   each   document  would  allow  for  a  finer  grain  analysis  at  a  regional  level.  In  addition,  one  of  the  main   challenges  of  any  social  media  analysis  is  to  understand  better  the  demographic  characteristics   of  the  sampled  population.  Finally,  online  conversations  are  a  relatively  new  source  of  data,  and   it   is   thus   required   to   start   maintaining   a   database   of   these   documents   over   time,   so   in   the   future,  analysis  will  cover  a  longer  period.   The   potential   is   high,   and   in   the   future,   models   that   are   more   robust   can   be   built   for   both   helping  predict  and  anticipate  the  consequences  of  unemployment  spikes,  and  real-­‐time  social   listening   could   be   used   to   monitor   the   experience   of   the   unemployment   in   an   automated   manner,  continuously.   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   12      
  • 13.       Annex  A:  Tools   Data  acquisition:  The  data  acquisition  is  handled  through  a  third  party  service,  Boardreader.   Data   filtering:   SAS®   Content   Categorization   software   is   used   to   create   the   filter   the   relevant   documents.  Data  must  satisfy  the  standards  set  forth  in  SAS  Content  Categorization  to  be  passed   to  the  next  step  in  the  process.   Dynamic-­‐Correlation   analysis:   Several   tools   have   been   used   for   the   analysis,   including   SAS®   Content  Categorization  Studio,  SAS®  9.2  statistical  and  forecasting  tools,  SAS®  Enterprise  Guide   and   SAS®   Sentiment   Studio   Workbench.   The   text   is   parsed   through   a   SAS   procedure   called   DOCPARSE.   The   SAS   procedure,   PROC   TIMESERIES,   is   used   to   obtain   the   cross-­‐correlation   function  results.   Mood   State:   Mood   State   is   a   method   by   which   SAS   measures   the   overall   mood   and   specific   moods  of  a  data  corpus.  Unlike  sentiment  analysis,  which  is  a  simple  positive/negative/neutral   decision,   mood   state   analysis   offers   a   more   refined   measure   by   which   to   judge   social   media.   Documents   are   scored   to   provide   mood   scores   for   Anxiety,   Confidence,   Hostility,   Confusion,   Energy,  and  Happiness.   Unemployment   dashboard:   The   data   and   results   are   represented   using   the   SAS   Business   Intelligence  Dashboard  Interface.         About  SAS   SAS   is   the   leader   in   business   analytics   software   and   services,   and   the   largest   independent   vendor  in  the  business  intelligence  market.  Through  innovative  solutions,  SAS  helps  customers   at   more   than   50,000   sites   improve   performance   and   deliver   value   by   making   better   decisions   faster.  Since  1976  SAS  has  been  giving  customers  around  the  world  THE  POWER  TO  KNOW®.  SAS   and   all   other   SAS   Institute   Inc.   product   or   service   names   are   registered   trademarks   or   trademarks   of   SAS   Institute   Inc.   in   the   USA   and   other   countries.   ®   indicates   USA   registration.   Other   brand   and   product   names   are   trademarks   of   their   respective   companies.   Copyright   ©   2011  SAS  Institute  Inc.  All  rights  reserved.     White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   13      
  • 14.       Annex  B   Example  of  documents  categorization  based  on  conversation  topics   Topic  Examples  (all  occur  within  same  document  as  mention  of  Unemployment  or  synonym)   • Alcohol   o “My  beer  budget  will  obviously  be  cut.”   • Borrow/Save  money   o “I  hate  to  do  it  but  will  have  to  get  a  payday  loan.”   • Consumer  goods   o “Definitely  won’t  be  buying  new  clothes  this  season.”   • Nutrition  quality   o “Not  sure  how  to  afford  fresh  produce  these  days.”   • Education   o “We  can’t  afford  the  tuition  so  she’ll  have  to  withdraw.”   • Underemployment   o “It’s  only  part  time  but  at  least  it’s  money  coming  in.”   • Entertainment   o “The  kids  will  make  do  without  new  video  games  or  dvds.”   • Financial   o “A  few  more  months  and  we’ll  have  to  seriously  consider  a  bankruptcy.”   • Bills   o “Sorry  water  bill,  this  month  I  pay  the  electric,  next  month  it’s  the  student   loans.”   • Healthcare   o “He  has  cobra  available  but  I  don’t  know  we’ll  ever  pay  for  it.”   • Housing   o Loss   § “That’s  it.  The  bank  is  foreclosing  next  week.”   o Downgrade   § “We’ll  save  a  bunch  if  we  can  move  to  a  smaller  place.”   • Transportation   o Loss   § “They’re  trying  to  repo  my  wife’s  car.”   o Downgrade   § “I  sold  the  Audi  and  got  a  used  Mazda.”   o Public   § “I  canceled  the  car  insurance  so  I’ll  start  taking  the  bus.  It’s  cheaper.”   • Travel   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   14      
  • 15.       o “The  Easter  vacation  we  planned  is  officially  canceled  due  to  lack  of  funds.”   • Unemployment  claim   o “I  filed  for  benefits  the  same  day.  Ughh.”   Example  of  documents  categorization  based  on  sentiment  analysis   • Anxiety   o “I’m  nervous  that  I  won’t  find  another  job  like  this  one.”   • Confidence   o “She  doubts  that  we  can  afford  to  have  two  cars.”   • Hostility   o “It’s  outrageous  that  we  were  only  given  two  weeks  severance!”   • Confusion   o “I’ve  felt  so  scattered  since  the  layoffs.”   • Energy   o “My  husband  has  been  so  sluggish  for  the  past  few  months.”   • Happiness   o “The  counselor  said  it  was  normal  to  experience  a  sense  of  mourning  after   losing  a  job.”   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   15      
  • 16.       Annex  C   Cross-­‐correlations  functions  between  mood  states  and/or  conversation  topics  against  the   official  unemployment  rate.   United  States     Figure  10:  (Hostile  Mood  –  Agreeable  Mood)  Hostility     Figure  11:  (Depressed  Mood  –  Elated  Mood)  Depression   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   16      
  • 17.         Figure  12:  (Unsure  Mood  –  Confident  Mood)  Uncertainty     Figure  13:  (Housing  Loss)   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   17      
  • 18.         Figure  14:  (Transportation  Repossession)   Ireland     Figure  15:  (Anxious  –  Composed  News)  Anxiety   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   18      
  • 19.         Figure  16:  (Confused  –  Clearheaded)  Confusion     Figure  17:  Confident  by  Blogs   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   19      
  • 20.         Figure  18:  Transportation  Downgrade     Figure  19:  Travel  Cancellations   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   20      
  • 21.         Figure  20:  Housing  Downgrade   White  Paper:  “Using  Social  Media  to  Add  Depth  to  Unemployment”   21