1. Visualizing Ontologies and Ontology Mappings May 17 th, 2012 (NIH/NCI Nano Working Group) Margaret-Anne Storey, Bo Fu, Lars Grammel, The CHISEL Group, University of Victoria 1

2. National Center for Biomedical Ontology Goal: develop innovative technology and methods that allow scientists to record, manage, and disseminate biomedical information and knowledge in both human readable and machine-processable form. the CHISEL group, University of Victoria 2

3. Our goal: Provide cognitive support for ontology developers and users through visual and intelligent user interfaces Visualizing ontologies: Diamond: Degree of interest browsing Jambalaya: visualizing ontologies FlexViz (currently in BioPortal) BioMixer (coming soon) Visualizing mappings and alignment: CogZ: Cognitive support for ontology alignment Visualizing mappings (coming soon)

4. What is cognitive support? A form of “mental assistance” during thinking and problem solving We often provide support through automation but real complexity is left to the user Distributed cognition

5. DIaMOND—Degree of Interest Modeling for Ontology Navigation and Development http://www.thechiselgroup.org/diamond

6. Motivation Navigating ontologies can be tedious… – Long scrolling lists, expanding/collapsing nodes – Large number of irrelevant elements occlude relevant information Users often don’t know where to start when navigating an unfamiliar ontology – Might appreciate “worn paths”

7. DIaMOND -- Approach Applies principles of attention-reactive interfaces – Mechanism to calculate user’s degree of interest (DOI) – Dynamic display of information using the DOI Goals – Draw user’s attention to interesting elements – Reduce navigation overhead

8. DIaMOND (Protégé plug-in) Three levels of interest – Interesting Has been interacted with such that the DOI value exceeds a threshold value – Uninteresting DOI value falls below the threshold value – Landmark: Hub concept Manually specified by user DOI value exceeds a threshold value Lightweight, easily reversible focus techniques Consistent with existing, familiar Protégé views

9. Highlighting and Filtering in the Class Browser: standard, highlighting, highlighting and filtering.

10. Diamond’s Future Work? Sharing DOI among users (many requests) Role and Task-based DOI calculations Use of instance data and annotated data to supplement DOI calculations Integrate and evaluate Diamond approach in BioPortal Develop heatmaps showing which concepts are navigated to and used

11. Jambalaya Visualization of ontologies to support navigation

12. Visualization and Jambalaya Why are visualizations important? – Ontologies can be extremely large and complex, difficult to navigate using just trees and lists – Visualizations help users understand ontologies by showing overviews and hub concepts Other visualization approaches: – OntoViz, OWLViz, TGViz, ezOWL, Ontorama, Ontobroker – IsaViz: visual RDF editor – Many graph-like visualization tools! – Indented lists, trees, landscapes, matrix, hyperbolic, 3D, treemap, zoomable, radar views etc.

13. Main Features in Jambalaya Classes and instances/individuals are represented by nodes in the graph Slots or properties are represented by arcs between nodes

14. the CHISEL group, University of Victoria

18. Inconsistencies “pop-out” in some views

19. Composite Arcs High level edges (lifted)

20. Different Views – Nested View (Default) – Flat views Class Tree Class & Instance Tree Domain & Range (OWL) – TreeMap View – Query View – Filmstrip – Customized views

21. FlexViz Some ideas from Jambalaya ported to a flash implementation and integrated with BioPortal Goals: – Reach a larger audience (web) – Promote socialization around the visualizations – Reduce task complexity (no install) – Track user behaviour (could inform Diamond-like features) – Exploratory interaction (as opposed to overviews) 21

22. FlexViz

23. Some FlexViz features Ontology tree synchronization with graph view Export graph as an image or an xml file, can be emailed as an attachment Example layouts - ForceDirectedLayout, SpringLayout, and IndentedTreeLayout FlexViz widget (embed within other websites)

24. But! Flexviz visualizations were cumbersome to customize Integration with other sites requires programming Used Flash Flexviz

25. Towards Visualization as a Service Enable the flexible visual exploration and analysis of biomedical ontologies and data Support collaboration in visual exploration and analysis of biomedical ontologies and data Enable presentation of analysis artifacts on the web

26. Visual analytics over the web… For example, IBM ManyEyes… NCBO Presentation March 2, 2011

27. BioMixer An online platform for the visual exploration of multiple biomedical ontologies 27

28. BioMixer vs. FlexViz: Motivation 28

29. BioMixer Architecture/Vision 29

30. Web-based Collaborative Workspaces 30

31. Multiple Coordinated Visualizations 31

32. Visualization Embeds iframe src="http://bio- mixer.appspot.com/?viewId=188001" width="480" height="400">Sorry, your browser doesn't support iFrames 32

33. BioPortal “precanned” Embeds: 33

34. Visualizing BioPortal usage data

35. Comparing some ontology usage across countries

36. Summary: Bio-Mixer Contributions Exploration of multiple ontologies and their mappings in a single graph visualization Multiple coordinated ontology visualizations Web-based and team enabled visual ontology exploration workspaces Authoring of visualization embeds for inclusion in BioPortal and external web content 36

37. Visualizing Mappings between Terms and between Ontologies 37

38. CogZ Cognitive Support and Visualization for Human- Guided Mapping Systems

39. Ontology alignment Mapping terms from one ontology to another, preserving structural relationships Research emphasis mostly on the algorithm Select ontologies Alignment algorithm Potential mappings Verified mappings Mapping complete

40. Ontology alignment continued Cognitively challenging The user must: – Make decisions about mapping candidates – Supply custom mappings that were missed – Understand the domains and structure of both ontologies – Remember decisions that were made Can be made easier through cognitive support

41. Visualization plug-in for Prompt Prompt: framework to support ontology alignment – Added a plug-in mechanism and perspectives to Prompt to support other algorithms and visualizations – CogZ: Integrating various visualizations and filters to support the user’s decision making process

42. the CHISEL group, University of Victoria

43. CogZ on the web (prototype only)

44. Multiple Ontologies and Mappings 44

45. Mappings between Terms - Matrix 45

46. Mappings between Ontologies 46

47. Next steps Integrating the matrix and graph based mapping views in BioMixer Appreciate feedback on these views! 47

48. Nanoparticle ontology University of Victoria48

49. University of Victoria49

50. Mappings between NPO and other ontologies

51. Visualization requirements for Nanoparticle ontology? Exploration, navigation, search? Support error checking? Changes across ontology versions? Publication and syndication of changes? Distinguish between original and imported classes/ontologies (upper versus domain ontologies)? Distinguish actual from inferred relations? Mappings? Support for alignment? Scalability? Is NPO stable in terms of size? Non graph views? e.g. charts? Collaboration features? Need for embeds? 51