1. 1 Helping online communities to semantically enrich folksonomies ISICIL, mai 2010 Freddy Limpens, Fabien Gandon Edelweiss, INRIA Sophia Antipolis {freddy.limpens, fabien.gandon}@inria.fr Michel Buffa Kewi, I3S, Univerité Nice-Sophia Antipolis Edelweiss

2. 2 How to turn folksonomies......into comprehensible topic structures ? ? pollution Soil pollutions has narrower pollutant Energy related

3. 3 … without overloading users

4. 4 … and by collecting all user's expertise into the process

5. 5 Our approach Integrate usage-analysis for a tailored solution Supporting diverging points of view Automatic processings + Human expertise through user-friendly interfaces

6. 6 Concrete scenario Experts produce docs + tag Archivists centralize + tag Public audience read + tag

7. 1. Modeling statements about tags

8. 8 Supporting diverging points of view car pollution skos:related

9. 9 car pollution John Paul Supporting diverging points of view agrees disagrees skos:related

10. Supporting diverging points of view 10 car pollution skos:related John Paul hasApproved hasRejected tagSemanticStatement (named graph)

11. Supporting diverging points of view

12. 2. folksonomy enrichment Life cycle

13. 13 Dataset DeliciousTheseNetCADIC What Bookmarks of users of tag "ademe" Keywords for Ademe's PhD projects Archivists indexing lexicon # tags 101565831439 # posts 1013 14254675 # (restricted tagging) 3015 1016025515 # users 812 14251

14. ADDING TAGS Automatic processing User-centric structuring User-centric structuring Detect conflicts Global structuring Global structuring Flat folksonom y Structured folksonom y Folksonomy enrichment life-cycle

15. 15 pollution pollutant pollution Soil pollutions 1. String-based metrics

16. 16 Evaluation of 30 edit distances Combining the best metrics Needs complement ! 1. String-based metrics

17. 17 1. String-based metrics Evaluated against Ademe expert's point of view

18. Related55,10% Broader/narrower32% CloseMatch12,82% Result on full dataset

19. Node size ↔ InDegree ◉ tags (delicious + thesenet) ◉ mot-clés svic

20. Result on full dataset Node size ↔ InDegree ◉ tags (delicious + thesenet) ◉ mot-clés svic

21. 21 Fig. Markines et al. (2009) Association via : Users tags 2. Tri-partite structure of folksonomies

22. tag1tag2tag3 tag1freq (tag1)cooc (tag1, tag2)cooc (tag1, tag3) tag2cooc (tag2, tag1)freq (tag2)cooc (tag2, tag3) tag3cooc (tag3, tag1)cooc (tag3, tag2)freq (tag3) 2. Tri-partite structure of folksonomies Tag-based association : => gives "related" relations

23. Example result on CADIC dataset:

24. 2. Tri-partite structure of folksonomies User-based association (Mika) : environnement agriculture U1 U1 U2 U2 U3 U3 U4 U4 U6 U6 => gives "subsumption" relations

25. Example result on Delicious dataset: Arrows mean "has broader" thickness ≈ weight

26. TheseNet dataset: Arrows mean "has broader" thickness ≈ weight

27. ADDING TAGS Automatic processing User-centric structuring User-centric structuring Detect conflicts Global structuring Global structuring Flat folksonom y Structured folksonom y Folksonomy enrichment life-cycle

28. 28 Embedding structuring tasks within everyday activity (searching e.g)

29. 29 Embedding structuring tasks within everyday activity (searching e.g)

30. 30 Capturing user's point of view

31. 31 Experimentation ADEME

32. ADDING TAGS Automatic processing User-centric structuring User-centric structuring Detect conflicts Global structuring Global structuring Flat folksonom y Structured folksonom y Folksonomy enrichment life-cycle

33. 33 Conflict detection environment pollution narrower broader

34. 34 Conflict detection environment pollution narrower broader Using rules e.g: IF num(narrower)/num(broader) ≥ c THEN narrower wins ELSE 'more generic' wins

35. 35 Conflict detection environment pollution narrower broader related broader narrower more generic

36. Ademe experimentation Total number of relations125 Contradictory43 Consensual14 Only rejected (to be deleted ?)2 Only proposed by computer (no user review)52 Debated (approved AND rejected at least once)14

37. Example result on Ademe's experimentation subset

39. ADDING TAGS Automatic processing User-centric structuring User-centric structuring Detect conflicts Global structuring Global structuring Flat folksonom y Structured folksonom y Folksonomy enrichment life-cycle

40. 40 environment pollution related ReferentUser Global structuring by Referent hasApproved

41. ADDING TAGS Automatic processing User-centric structuring User-centric structuring Detect conflicts Global structuring Global structuring Flat folksonom y Structured folksonom y Folksonomy enrichment life-cycle

43. 43 Take away message (conclusion)

44. 44 Help communities structure their tags What we do :

45. 45 Our contributions: Usages analysis Automatic processing of tags Tag structuring embedded in every-day tools Supporting multi-points of view

46. 46 Future work Interfaces : to capture user-centric contributions Global administrations (Referent User) Tag searching Real-scale experimentation Mapping between knowledge representation (Gemet thesaurus – Tags – CADIC e.g.) Moving to other socio-structural models

47. 47 Thank you ! freddy.limpens@inria.fr http://www-sop.inria.fr/members/Freddy.Limpens/ http://isicil.inria.fr

48. 48

49. 49 What is a tag ?

50. 50 Tagging model

51. 51 Tagging model

52. 52 Hypotheses Tag-link is thematic Tags are concept- candidate

53. 53 Supporting diverging points of view

54. 54 An eco-system of agents

55. 55

56. 56 Tags are nice to organize your own resources...

57. 57... but also to get involved in the organization of shared resources