Improving decision-making based on government data and visualizations
Upcoming SlideShare
Loading in...5
×
 

Improving decision-making based on government data and visualizations

on

  • 397 views

 

Statistics

Views

Total Views
397
Views on SlideShare
397
Embed Views
0

Actions

Likes
0
Downloads
5
Comments
0

0 Embeds 0

No embeds

Accessibility

Categories

Upload Details

Uploaded via as Apple Keynote

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n

Improving decision-making based on government data and visualizations Improving decision-making based on government data and visualizations Presentation Transcript

  • Improving decision makingbased on government data and visualizations Alvaro Graves gravea3@rpi.edu 1
  • Agenda• Background • Open Government Data• Problem • How to use this data?• Proposed Solution • Personas • (Re)use of visualizations• Future Work 2
  • Background 3
  • Open Government Data• Governments are releasing huge amounts of data (geographical, budget, transit, etc)• Goal: Improve transparency, economy, make people take informed decisions, etc.Open data is the electricity of the 21st century! - M. Hausenblas 4
  • The government data landscape• Independent Data• Different goals Consumer (In govt)• No coordination Data Data Civil• Highly decoupled Hacker• Asynchronous Data Data Data Data Producer Data Data Consumer (data journalist) 5
  • ScenarioProblem: Some stakeholders can’t use most ofthis government data and use them in theirdecision-making process, since don’t have theskills or training needed to consume it.**Based on interviews 6
  • Objectives• Our goal: Allow more people to use and understand government data to make more informed decisions• A solution: Improve creation, sharing and reuse of data-based visualizations, so they can consume and communicate data 7
  • Challenges• Who are the stakeholders? • Govt. Data producers and consumers, Data journalists/ Activists, Civil hackers, Citizens• How do we help people to (re)use all this data? • Use of visualizations as a medium to communication [1] • ... but this is hard [2] • How can we ease these processes? [1] Crapo, A.W., et al. Visualization and the process of modeling: a cognitive-theoretic view, 2000 [2] Viegas, F.B., et al. Manyeyes: a site for visualization at internet scale, 2007 8
  • Who are thestakeholders? 9
  • Stakeholders• Government Data Provider• Government Data Consumer• Data Journalist / Activist• Civil Hacker • Already use the data, have the skills• Common Citizen • Not interested [3] [4] in being part of this ecosystem (directly)[3] DiFranzo, D. and Graves, A. A Farm in Every Window: A Study into the Incentives for Participation inthe Windowfarm Virtual Community, 2010[4] Preece, J. and Shneiderman, B. The reader-to-leader framework: Motivating technology-mediated 10social participation, 2009
  • Profile modelling using Personas• Personas[5] is a technique common in HCI and human factors to understand user types• Based on interviews, create a “persona” that represents a set of users with common characteristics• Add as much many details as possible to understand environment, [5] Blomkvist, S. Personas - An overview, 2004 11
  • Persona: Government data provider*• Phillip Mancini, 35, married, one daughter.• He is a data analyst working for the agency for Electronic Government• His work consists in promoting the government’s data portal • This means coordinate and request data from other agencies and publish it in the government portal • Promote and make easier for others to use the data available • He knows some programming, but he is not an expert (he knows well several datasets though) • Eventually create mashups to his boss or other government employees to show the benefits of Open Data (but he doesn’t have much time/expertise for this)* Based on interviews with government employees 12
  • How can we helppeople to (re)use all these data? 13
  • Visualizations as a way to consume and share data • Visualizations are a simple way for humans to communicate data and quantitative information[6] • A visualization can be • A graph • Full Chart Title Goes Here Subtitle appears here if it exists Pie 1 5 1 2 Y Axis Label Category A Category B Category C Category D • X Axis Label Scatterplot • Full Chart Title Goes Here Subtitle appears here if it exists Others 15 12 Y Axis Label 9 6 3 • Category A Category B Category C Category D X Axis Label A table, list • A map [6] Few, S. Data Visualization for Human Perception, Encyclopedia of Human Interaction, 2010 14
  • Problems for the creator*• Create visualizations is hard • Creator needs to understand underlying data • Creator needs to choose a visualization strategy• Visualizations of Open Government Data • Different formats • Distributed data • Focus on how to tie everything up* Based on preliminary interviews (Govt. data provider & consumer) 15
  • Problems for the observer*• Accountability questions• Visualization’s provenance • Where does the data Full Chart Title Goes Here Subtitle appears here if it exists 15 12 Y Axis Label 9 comes from? 6 3 Category A Category B Category C Category D X Axis Label • When was collected? • How was processed? *Based on preliminary interviews (Data journalist) 16
  • Problems for the reuser*• “I wonder how this data looks in a map”• “What if we use the data from previous year?”• “What if we take the median instead of the average?” Full Graph Title Goes Here Subtitle appears here if it exists 15 12 Y Axis Label 9 6 3 0 10 20 30 40 50 60 X Axis Label* Based on interviews (Govt. data consumer & Data journalist) 17
  • How can we ease theprocess of creating andreusing a visualization? 18
  • Visualizations as declarative components• Instead of forcing users to interact with code, use formal components that mediates between a user and the computer• This components will reduce the efforts, training and skills Full Chart Title Goes Here Subtitle appears here if it exists necessary to create 1 5 1 Y Axis Label 2 9 visualizations 6 3 Category Category B Category C Category D A X Axis Label 19
  • Step 1: Encode this knowledge• Use of semantics opmv:Process opmv:used opmv:Artifact cnt:ContentAsText opmv:wasControlledBy skos:Concept opmv:Agent NameId rdfs:subClassOf rdfs:subClassOf rdfs:subPropertyOf rdfs:subClassOf rdfs:label rdf:type :Application :Message rdfs:subPropertyOf dc:hasFormat to represent rdfs:subClassOf skos:broader blank :usedParameter :Component :usedInput cnt:chars rdfs:subClassOf dc:format rdfs:subClassOf rdfs:subClassOf rdfs:subClassOf Code rdfs:subClassOf :VisualizationComponent :DataComponent• High-level :ProcessComponent mime type :Input :Parameter rdfs:subClassOf rdfs:subClassOf :UrlDereferencer :SparqlEndpointRetriever representation of different Full Graph Title Goes Here component of a Subtitle appears here if it exists 15 12 Y Axis Label 9 6 3 visualization 0 10 20 30 40 50 60 X Axis Label opmv:A opmv:Proc opmv:Arti cnt:ContentAs skos:Con opmv:wasControlledBy opmv:used gent ess fact Text cept NameId rdfs:subClassOf rdfs:subClassOf rdfs:subPropertyOf rdfs:subClassOf rdfs:label :Applicati rdf:type on :Mess rdfs:subPropertyOf age dc:hasFormat rdfs:subClassOf skos:broader blank :usedParameter :Compon ent :usedInput cnt:chars rdfs:subClassOf dc:format rdfs:subClassOf rdfs:subClassOf rdfs:subClassOf :VisualizationCompo Code rdfs:subClassOf :DataCompo nent nent :ProcessCompone nt mime :Input :Parameter type rdfs:subClassOf rdfs:subClassOf :UrlDereference :SparqlEndpointRetr r iever <HTML> 20
  • Step 2: Explore Visualization• Allow users to obtain the formalization of it • High-level Full Graph Title Goes Here Subtitle appears here if it exists 15 skos:Concept opmv:Agent opmv:wasControlledBy opmv:Process opmv:used opmv:Artifact NameId cnt:ContentAsText components rdfs:subClassOf 12 rdfs:subClassOf Y Axis Label rdfs:subPropertyOf rdfs:subClassOf 9 rdfs:label rdf:type 6 :Application :Message 3 rdfs:subPropertyOf dc:hasFormat rdfs:subClassOf skos:broader 0 10 20 30 40 50 60 blank :usedParameter X Axis Label :Component :usedInput cnt:chars rdfs:subClassOf dc:format rdfs:subClassOf rdfs:subClassOf rdfs:subClassOf opmv:Proc opmv:Arti cnt:ContentAs Code skos:Con opmv:A opmv:wasControlledBy opmv:used gent ess fact Text rdfs:subClassOf :VisualizationComponent cept NameId rdfs:subClassOf rdfs:subClassOf rdfs:subPropertyOf rdfs:subClassOf rdfs:label :DataComponent • The relations :Applicati rdf:type on :Mess rdfs:subPropertyOf age dc:hasFormat rdfs:subClassOf skos:broader blank :ProcessComponent :usedParameter :Compon ent :usedInput cnt:chars mime type rdfs:subClassOf rdfs:subClassOf rdfs:subClassOf dc:format :Input :Parameter rdfs:subClassOf :VisualizationCompo Code rdfs:subClassOf :DataCompo nent nent :ProcessCompone nt rdfs:subClassOf rdfs:subClassOf mime :Input :Parameter type rdfs:subClassOf rdfs:subClassOf :UrlDereference :SparqlEndpointRetr r iever :UrlDereferencer :SparqlEndpointRetriever <HTML> among them• Display it in graphical terms (workflow, forms, etc) 21
  • Step 3: Reuse of a visualization• Modify a new copy of a visualization • Represented as a formalization to the user, no code Full Graph Title Goes Here Full Chart Title Goes Here Subtitle appears here if it exists Subtitle appears here if it exists 15 15 12 12 Y Axis Label Y Axis Label 9 9 6 6 3 3 Category A Category B Category C Category D 0 10 20 30 40 50 60 X Axis Label X Axis Label opmv:wasControlledBy opmv:Process opmv:used opmv:Artifact cnt:ContentAsText skos:Concept opmv:Agent opmv:wasControlledBy opmv:Process opmv:used opmv:Artifact cnt:ContentAsText NameId skos:Concept opmv:Agent rdfs:subClassOf NameId rdfs:subClassOf rdfs:subPropertyOf rdfs:subClassOf rdfs:subClassOf rdfs:subClassOf rdfs:label rdfs:subPropertyOf rdfs:subClassOf rdf:type :Application rdfs:label :Message rdf:type :Application rdfs:subPropertyOf dc:hasFormat :Message rdfs:subClassOf skos:broader rdfs:subPropertyOf blank dc:hasFormat :usedParameter rdfs:subClassOf skos:broader blank :usedParameter :Component :usedInput cnt:chars :Component :usedInput cnt:chars rdfs:subClassOf dc:format rdfs:subClassOf rdfs:subClassOf rdfs:subClassOf rdfs:subClassOf Code dc:format rdfs:subClassOf :VisualizationComponent rdfs:subClassOf rdfs:subClassOf rdfs:subClassOf Code :DataComponent rdfs:subClassOf :VisualizationComponent :DataComponent :ProcessComponent :Input :Parameter mime type :ProcessComponent :Input :Parameter mime type rdfs:subClassOf rdfs:subClassOf rdfs:subClassOf rdfs:subClassOf :UrlDereferencer :SparqlEndpointRetriever :UrlDereferencer :SparqlEndpointRetriever <HTML> <HTML> Backlinking 22
  • What should we measure?• Time required to complete tasks • Create visualization from scratch vs. using formalization • Reuse visualization from scratch vs. using formalization• Self report • Can you do a task you weren’t able to do before? • Can you perform better (time, # errors) using this approach? 23
  • Future work• Do a more complete creation of personas • Work with more Data Producers and Data Journalists• Build tools based on our formalization • Several components already created• Test it against real users • Design experiments in details • A dozen volunteers available so far 24
  • References• [1] Crapo, A.W., et al. Visualization and the process of modeling: a cognitive- theoretic view, 2000• [2] Viegas, F.B., et al. Manyeyes: a site for visualization at internet scale, 2007• [3] DiFranzo, D. and Graves, A. A Farm in Every Window: A Study into the Incentives for Participation in the Windowfarm Virtual Community, 2010• [4] Preece, J. and Shneiderman, B. The reader-to-leader framework: Motivating technology-mediated social participation, 2009• [5] Blomkvist, S. Personas - An overview, 2004• [6] Few, S. Data Visualization for Human Perception, Encyclopedia of Human Interaction, 2010 25