Creation of visualizations based on Linked Data
3 likes1,648 views
The document discusses the creation of visualizations for linked data, addressing challenges faced by non-technical users and domain experts. It introduces 'Visualbox', a tool designed to simplify the visualization process while emphasizing the importance of query creation. User studies revealed that while Visualbox facilitates visualization, the complexity of SPARQL query generation remains a significant barrier.
Creation of visualizations based on Linked Data
- 1. Creation of visualizations based on Linked Data Alvaro Graves gravea3@rpi.edu - @alvarograves Tetherless World Constellation Rensselaer Polytechnic Institute 1 Thursday, June 20, 13
- 2. Agenda • Challenges in Linked Data usage • How to create an LOD visualization? • User study • Conclusions 2 Thursday, June 20, 13
- 3. Challenges in Linked Data usage 3 Thursday, June 20, 13
- 4. Good and bad news about Linked Data • Good: • Billions of triples available on the Web • Lots of information about people, events, places, etc. • Not so good: • Not a lot people outside the Semantic Web community find it usable 4 Thursday, June 20, 13
- 5. Who are the users? • People with domain expertise, not technical • Journalists • Scientists (Bio-, Chemo-, X-Informaticians) • Novice Semantic Web students • People who want an easy/rapid way to understand certain aspects of LOD datasets 5 Thursday, June 20, 13
- 6. Challenges • What is needed for non-technical users to consume and explore Linked Data? • How can we empower domain experts that lack experience with Linked Data? 6 Idea: Use of visualizations can simplify consumption of Linked Data Thursday, June 20, 13
- 7. WhyVisualizations? • Visual inspection of data is helpful • Can help to detect trends and outliers • It is easy to share and consume for non data experts 7 Anscombe’s quartet “Attractions in Berlin” Thursday, June 20, 13
- 8. How to create an LOD visualization? 8 Thursday, June 20, 13
- 9. Visualization workflow 9 2) Visualization step 3) Sharing step (optional) 1) Query step Query writing Query testing Creationg of visualization Sharing Thursday, June 20, 13
- 10. Visualbox • “Easy Linked Data visualizations” • Integrated environment to create visualizations based on LOD • Makes use of available SPARQL endpoints • Provides 14 types of visualizations out-of-the- box 10 Thursday, June 20, 13
- 11. 1 - Existing visualizations 2 - Query editor 3 -Visualization Editor 4 -Visualization helper Visualbox 11 5 - Buttons for query testing 6 - Query testing results 7 - Sharing buttons Thursday, June 20, 13
- 12. Mozilla Festival • TestedVisualbox against real users • Most of them never heard about Linked Data 12 Thursday, June 20, 13
- 13. Feedback from Mozilla Festival • Positive: Most people evaluated it as a great experience for learning about Linked Data and visualizations • Negative: Most people found hard to even start working with SPARQL • Unsure about the syntax • Unsure “what’s in there” in an endpoint 13 Thursday, June 20, 13
- 14. User study 14 Thursday, June 20, 13
- 15. User study • People with basic to medium knowledge of SPARQL and Linked Data • Participants watched an introductory video • Used think aloud approach • Three tasks, visualizations based on DBpedia • Ex-post subjective evaluation 15 Thursday, June 20, 13
- 16. Results • Most of the time (5-10 min.) was focused on creating and debugging SPARQL query • Once the query was ready, the rest of the process was very short (1-2 min.) 16 Thursday, June 20, 13
- 17. Participant’s feedback • Positive: Visualbox makes it easy to create visualizations • Much easier than doing it from scratch • Add features to power users • Negative: Creation of SPARQL query • One size does not fit all 17 Thursday, June 20, 13
- 18. Possible solutions • Syntax assistance • Namespaces and prefixes • Improved syntax highlighting • Recommendations/suggestions based on: • Endpoint’s content • What others have queried • Other existing visualizations 18 Thursday, June 20, 13
- 19. Conclusions • Creating LOD visualizations is a two-step process: • Query creation • Visualization development • Visualbox provides an integrated environment for the first step • It reduces the effort for the second step • There are open challenges in helping users to create SPARQL queries 19http://visualbox.org - gravea3@rpi.edu - @alvarograves Thursday, June 20, 13