Building Collaborative Knowledge Representations in Real Time

1. Verizon500 Westchester Ave. White Plains, NY USA 10604 albert.m.selvin@verizon.com Knowledge Media Institute Open University Milton Keynes, UK MK7 6AA Building Collaborative Knowledge Representations in Real Time An Analysis of Facilitative Micro-ActionsInterim Progress ReportAl Selvin

2. Agenda • Goals • Research questions • Context • Method of analysis • Results to date • Summary and next steps

3. The Vision Thing Bring together the technical side of knowledge engineering with performative and creative aspects from the arts and humanities. Bring the benefits and capabilities of artistry to the practice of creating semi-formal knowledge representations with groups in real time.

4. More Vision • The consequences of this union can be to address some long-standing problems in knowledge technology • Effective long-term organizational memory • Overcoming the capture bottleneck • Enabling knowledge codification and formalization on the fly • “Value Now and Value Later”

5. Goals • Expanding the use, and usefulness, of collaborative knowledge media • Overcoming obstacles to adoption • Understanding aesthetic and ethical issues in the use of such technologies for groups and teams

6. Research questions • What is expert human performance in creating and modifying knowledge representations for groups, on the fly • Develop a descriptive vocabulary of sufficient granularity and nuance • What skills are actually used by expert practitioners • How can an understanding of the above contribute to identifying • Needed skills • Training methods • Improved software support • Increased adoption

7. Context for this analysis • NASA Mobile Agents field trial • Supporting RST scientists distributed in multiple locations in • Analyzing incoming science data • Formulating recommendations to the hab crew re specific goals, objectives, and tasks • Improving RST/crew processes in general, and • Understanding and improving the role of collaboration software tools and strategies in particular • Before, during, and after team meetings/telecons (SOWGs)

8. Roles and responsibilities • RST members • Review materials beforehand • Prepare analyses • Participate in RST telecons • RST lead • Chair the telecons • Crew uplink lead • Create knowledge representations • Publish materials from crew sessions • Crew members • Participate in crew sessions • Create knowledge representations (using Compendium) • Meeting Replay team • Create web-based videos of crew sessions integrated with Compendium knowledge representations • Science Organizer team • Integrate and maintain SO repository of science data

9. Roles and responsibilities • RST facilitator • Gather, prepare, and publish materials before RST meetings • From Science Organizer, Meeting Replay, Crew Compendium exports, RST analyses, emails, and other resources • Arrange telecon/web conferences • Convene sessions • Assist in locating and analyzing science data • Capture discussion and decisions during the sessions • Assist RST with software/tool issues • Build and modify Compendium knowledge representation on the fly • Retrieve materials from other tools and repositories and integrate them into the knowledge representation • Create summary materials at the conclusion of each session • Publish the materials to the web and other repositories

10. Configuration

11. Methodology maps Science Organizer item Hab crew maps Web resources Create Portal map Images Emails RST Analyses Publish to web Convene Webex session

12. 11m31s Listening to the participants, creating nodes, choosing node types, typingsummary comments and observations

13. 11m58s Capturing deliberation in nodes

14. 12m40s Creating logical ‘containers’ and drawing semantic links betweenconcepts

15. 26m23s Adding annotations

16. 36m22s Tagging key nodes with metadatato aid later recall and reuse

17. 121m04s Working with the RST to locate, analyze, cross-reference, and raise issues about the science data

18. 128m59s Create final maps for web export, harvesting nodes from earlier in the session and mapping them onto pre-made templates of summary questions

19. Example

20. Expertise required • To perform RST Facilitator role: • Listening and interpreting • Intervening in ‘normal’ conversation flow • Getting validation for captured material • Building hypertext representations on the fly • Interrelating data and objects • Adding metadata • Software-specific skills Conventionalfacilitationskills Knowledgemediafacilitationskills

21. Analysis method • Grounded theory (Strauss and Corbin) • Close analysis of a session, paying special attention to • Participant statements • Practitioner actions • Practitioner statements • Compendium moves • Building up explanatory concepts, categories, and properties • Focus on the engagement of both practitioner and participants with the Compendium representation

22. Emerging categories and concepts • Participant map engagement • The way in which participants relate to the current move • 4 types, 3 subtypes • Active (Text, Structure, Navigation) • Direct • Partial/Unclear • Delinked

23. Emerging categories and concepts • Compendium moves • 50 types, 44 subtypes • 646 individual moves in the analyzed session

24. Emerging categories and concepts • Practitioner verbal moves • 5 types • Statement/Announcement • Acknowledgement • Query • Helpful Comment • Exclamation • 146 individual moves in the analyzed session

25. Emerging categories and concepts • Activity types • 13 types, 9 subtypes • The primary type ofactivity the practitioneris engaged in • Can be as short as onemove or span many minutes

26. Emerging categories and concepts • Practitioner stances • The position of the practitioner with regard to the current activity • 5 types • Knowledge Navigator • Facilitator • Participant • Technical Expert • Editor

27. Emerging categories and concepts • Compound moves • Collections of individual moves that accomplish a simple action (e.g. navigating to a map, copying a node, navigating back to another and pasting it) • Mini-projects • Sequences of moves that accomplish a more complicated action with a clear goal • I’ve not yet created types of compound moves or mini-projects

28. Emerging categories and concepts • Practitioner response/engagement mode • The way in which the practitioner is engaging with the participants, on a move-by-move level • 4 types • Delinked • Indirect • Semi-direct • Direct

29. Emerging categories and concepts • Focus of moves • The objects with which the practitioner is engaging in a move • Can be multiple • 6 types • Participants • Maps • Text • Subject matter • Surroundings • Process

30. Emerging categories and concepts • Notes • Field notes, observations or commentary about particular moves, mini-projects, statements, or episodes • Themes • The stories that the participants and practitioner weave around the ostensible agenda items and formal discussion • In this session, 8 principal themes

31. Analysis spreadsheet

32. Analysis spreadsheet Activity types andpractitionerstance Practitioneractions Compendiummoves Timestamp Participantstatements Practitionerstatements Participantmap engagement

33. Analysis spreadsheet Engagementmode Compoundmoves Mini-projects

34. Analysis spreadsheet Fieldnotes Focustargets Themes / stories

35. Results to date: Practitioner stance

36. Results to date: Activity types

37. Results to date: Focus of practitioner moves

38. Results to date: Participant map engagement

39. Results to date: Practitioner engagement mode

40. Results to date: Practitioner verbal moves

41. Summary • Beginning to feel that it is possible to construct a useful descriptive vocabulary beyond the usual glosses (“discussion capture” “meeting facilitation” etc.) • Categories and concepts in an early stage of development • Need better ways to visualize results • Eager to apply the preliminary framework to other settings and practitioners

42. Next steps • Continue to refine the descriptive framework • Analyze Meeting Replay session of hab crew facilitation • Analyze working session between two skilled practitioners • Seek out green field locale