An Embeddable Dashboard for Widget-Based Visual Analytics on Scientific Communities
•
3 likes•2,284 views
Presented at i-KNOW 2012, the 12th International Conference on Knowledge Management and Knowledge Technologies, 5 September, Graz, Austria
Report
Share
An Embeddable Dashboard for Widget-Based Visual Analytics on Scientific Communities
- 1. 12th International Conference on Knowledge Management and Knowledge Technologies (i-KNOW 2012) 5-7 September 2012 Graz, Austria An Embeddable Dashboard for Widget-Based Visual Analytics on Scientific Communities Michael Derntl, Stephan Erdtmann, Ralf Klamma RWTH Aachen University Advanced Community Information Systems (ACIS) Informatik 5, Aachen, Germany derntl@dbis.rwth-aachen.de Lehrstuhl Informatik 5 (Information Systems) Prof. Dr. M. Jarke 1 These presentation slides by Michael Derntl are licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.
- 2. Motivation Scientific results socially created in scientific communities1 Quality of products success of community Stakeholder interest in success factors Classic indices: publications, citations – established communities New publication channels, new collaboration means – Web 2.0, social media/networks, etc. – Scattered information and large data volumes Lehrstuhl Informatik 5 (Information Systems) Prof. Dr. M. Jarke 1W. A. Kornfeld and C. Hewitt. The Scientic Community Metaphor. IEEE Trans. Syst., Man, and Cybern., SMC-11(1):24-33, 1981. 2
- 3. Data Visualization Data visualization eases assimilation of information Modern web technologies enable interactive visualizations © Google, Inc. Popular metaphor: dashboards Lehrstuhl Informatik 5 – One page overview, aggregate data, focus on important (Information Systems) Prof. Dr. M. Jarke 3 data, interactive components
- 4. Dashboards and Visual Analytics Requirements for visual analytics1 – Interactive display supporting exploration and reasoning – Stakeholder oriented flexible view – personalization, … Dashboard: viable tool for visual analytics on scientific community datasets? http://cdn.speedhunters.com/wp- Lehrstuhl Informatik 5 content/uploads/2010/01/gasser7.jpg (Information Systems) Prof. Dr. M. Jarke 1J. J. Thomas and K. A. Cook, eds. Illuminating the Path: The Research and Development Agenda for Visual Analytics. IEEE, 2005. 4
- 5. Pilot Community: TEL TEL-Map: roadmapping and weak signal analysis in Technology Enhanced Learning (TEL); supported by the TEL Mediabase1 Lehrstuhl Informatik 5 (Information Systems) Prof. Dr. M. Jarke 1M. Derntl, R. Klamma. A Mediabase for Technology Enhanced Learning in Europe. IEEE Learn. Tech. Newsletter, 14(3):2-5, 2012. 5
- 6. Challenges Configuring personalized dashboards – Addressing different stakeholder groups – Easy to set up / configure / use for non-technical people – Reusability of the solution beyond the pilot case Interactive visualization of arbitrary database query results – Different types of charts / graphs – New visualizations should easily be createable – Compatible with existing and new databases Lehrstuhl Informatik 5 (Information Systems) Prof. Dr. M. Jarke 6
- 7. Objectives Web-based dashboard for visualising SQL query results Decoupling the hosting application from dashboard application Implementation of – Dashboard frame containing visualization widgets – Data visualization generator app For pilot case at learningfrontiers.eu: – Dashboard frame as a Drupal module Lehrstuhl Informatik 5 (Information Systems) Prof. Dr. M. Jarke – Reuse of user credendials 7
- 8. Use Cases & GUI Embedding Application Widget Container Control Viz 1 Viz 2 Pane Viz n Lehrstuhl Informatik 5 (Information Systems) Prof. Dr. M. Jarke 8
- 9. Lehrstuhl Informatik 5 (Information Systems) Prof. Dr. M. Jarke http://learningfrontiers.eu/?q=dashboard 9
- 10. Widget Creator Selecting a database connection Entering an SQL query Selecting a visualization type Instant preview using Google Visualization API Publish as OpenSocial Lehrstuhl Informatik 5 (Information Systems) Prof. Dr. M. Jarke widget 10
- 11. Technical Architecture Dashboard Legend Control Pane Widget Container Previously Visualization existing Layer Visualization Widgets Newly implemented LAS Ajax Client Lightweight Application Server (LAS)1 LAS HTTP Connector Application Layer LAS Services LAS Database Connector Data User & Widget Data Sources for Data Sources for Lehrstuhl Informatik 5 Layer Data Visualization Visualization (Information Systems) Prof. Dr. M. Jarke 1M. Spaniol et al. LAS: A Lightweight Application Server…In Proc. i-KNOW 2006, pp. 592-599, 2006. 11
- 12. Components Interaction Dashboard Container Lightweight Application Application Data Legend 2 1a Server (LAS) Component of the Dashboard dashboard framework User Data User Management 1b 3 Component of the Service embedding application 6b Widget Data 4a Dashboard 4b 5a 6a Service Visualization Data Sources Widget 5b Query 5c Database(s) Service Visualization Layer Application Layer Data Layer 1 Register user (on first visit; automatically done by the embedding application) 2 Hand over LAS user credentials to the dashboard container 3 Dashboard container log in user on LAS 4 Retrieve list of available visualization widgets 5 Display visualization in widget 6 Store user preferences (after all personalization actions; during evaluation: action log) Lehrstuhl Informatik 5 (Information Systems) Prof. Dr. M. Jarke 12
- 13. End-User Evaluation Goal – Evaluation of usability and usefulness – Obtain feedback for improvement Methodology – Creation of ~20 visualization widgets on 3 TEL-Map databases – Definition of default dashboard for first login – TEL-Map partners / researchers tested according to predefined tasks (=> use cases) – Rating of the test participants has been determined using surveys (18 five-point Likert scale items + 2 open questions) Lehrstuhl Informatik 5 (Information Systems) – Collection of usage data (user action log) Prof. Dr. M. Jarke 13
- 14. Evaluation: Participants Survey was completed by 20 participants from TEL context On average they – dedicate 2/3 of their current working time to TEL. – have 8.5 years of experience with TEL Lehrstuhl Informatik 5 (Information Systems) Prof. Dr. M. Jarke 14
- 15. Evaluation: Usability Results Higher scores for usability compared to usefulness Negative correlation (p<.01) between years of experience and usability/usefulness N = 20 (M = 4.21 ± .68) 1 2 3 4 5 It is easy to remove widgets 4.90 It is easy to rearrange widgets 4.40 It is easy to resize the widget height and width 4.15 It is easy to detach widgets 4.55 Usability It is easy to find the description of a widget 4.15 The widget descriptions are sufficient to understand the 3.85 widget content The search interface is easy to use 3.85 It was easy to find relevant widgets 3.50 Lehrstuhl Informatik 5 (Information Systems) It was easy to add available widgets 4.55 Prof. Dr. M. Jarke 15
- 16. Evaluation: Usefulness Results Higher scores for usability compared to usefulness Negative correlation (p<.01) between years of experience and usability/usefulness N = 20 (M = 3.72 ± .84) 1 2 3 4 5 The data covered by available widgets appears relevant 3.80 to me The data covered by the widgets appears relevant to the 4.05 TEL community The available widgets cover a broad range of TEL 3.90 The available categories are useful to search for 3.45 widgets Usefulness The style of the data visualizations is appropriate 3.65 The range of available data visualizations is sufficient 3.40 This dashboard has all the functions and capabilities I 3.05 expect it to have The dashboard is useful to support mapping and 4.11 Lehrstuhl Informatik 5 roadmapping activities in TEL (Information Systems) The dashboard can support different TEL stakeholders 4.00 Prof. Dr. M. Jarke in observing data 16
- 17. Evaluation: Qualitative Results Experiences with creating a new widget (exploring the data, offering visualizations) – 14 tried, 7 succeeded (3 w/ error message on embed) – Problems: Incomplete documentation, DB schemas unknown Suggestions for improvement – Missing access to data behind a visualization – Useful for statistics about TEL, but less useful for understanding the future – Problems with meaning of visualizations and categories – Complex visualizations hard to read – Data export functionalities needed Lehrstuhl Informatik 5 (Information Systems) Prof. Dr. M. Jarke – Unintuitive interaction with the dashboard 17
- 18. Evaluation: Action Log Action Ranking 687 actions by 25 users traced 0 50 100 150 200 250 300 over 13 days (~27 per user) Toggle control pane 196 Embed widget 141 Remove widget 101 Move widget 88 Top 10: # Embeds / Widget / Category Search by category 50 0 1 2 3 4 Search by keyword 30 Resize column 26 Search (1) 4.00 Resize widget 28 Social (1) 4.00 Create widget 17 During evaluation Radar* (3) 3.00 Add/Rem column 10 After evaluation Network (4) 2.75 Content (3) 2.33 Histogram* (4) 2.25 Papers (7) 2.14 Blogs (6) 2.00 Geographical (6) 2.00 Lehrstuhl Informatik 5 (Information Systems) Pie* (4) 2.00 Prof. Dr. M. Jarke 18
- 19. Conclusions and Further Work Key points – Dashboard embeddable anywhere with few HTML lines – Visualization of scientific community data – OpenSocial based widgets – Google Visualization API based on SQL query results – Arbitrary relational DB sources possible – Pilot case: Drupal-based TEL portal; generally positive evaluation Limitations – Visualizations limited to single SQL query – More interactive and explorative features needed for visual Lehrstuhl Informatik 5 analytics tasks (Information Systems) Prof. Dr. M. Jarke 19