1. U NIVERSITY OF M INNESOTA Altruism, Selfishness, and Destructiveness on the Social Web GroupLens Research University of Minnesota John Riedl

2. U NIVERSITY OF M INNESOTA Bowling Alone (Amazon reviews)

3. U NIVERSITY OF M INNESOTA

4. Adaptive Hypermedia 2008 4 Tags scale: Library of Congress: 20M books in 200 years. www.librarything.com: 22M books in 3 years. Tag draw relevance from “the wisdom of crowds” Tags scale: Library of Congress: 20M books in 200 years. www.librarything.com: 22M books in 3 years. Tag draw relevance from “the wisdom of crowds”

5. Adaptive Hypermedia 2008 5 Messages Community-maintained Artifacts of Lasting Value o Requires User Modeling and Adaptive Hypermedia Key Research Challenges: o Attract contributions o Maintain quality o Achieve agreement

6. Adaptive Hypermedia 2008 6 Alexa Germany

7. U NIVERSITY OF M INNESOTA 1. Google (German) 3. Google (English) Search

8. Adaptive Hypermedia 2008 8

9. 9 Google PageRank Value of a page is the value of the pages that link to it Recursive! Algorithms and Psychology The Rich get Richer

10. Adaptive Hypermedia 2008 10 Web Structure

11. U NIVERSITY OF M INNESOTA (Web Search) shared Maurice Coyle and Barry Smyth AH’08

12. Adaptive Hypermedia 2008 12 Research Questions How can we mine free activity? What are the risks in these data?

13. U NIVERSITY OF M INNESOTA 2. YouTube Video by Amateurs

14. Adaptive Hypermedia 2008 14 Chocolate Rain by Tay Zonday Adam Bahner, a Ph.D. student in American Studies at the University of Minnesota Number 2 hottest viral video in history o Hottest viral video of Summer 2007 o Over 26 million views

15. Adaptive Hypermedia 2008 15 Videos Life Fast, Die Young

16. Adaptive Hypermedia 2008 16

17. Adaptive Hypermedia 2008 17 Huberman Dynamics of Viral Marketing The Dynamics of Viral Marketing, ACM TWeb 2007, Leskovec et al., HP

18. Adaptive Hypermedia 2008 18 Maximizing the Spread of Influence through a Social Network, David Kempe, Jon Kleinberg, Éva Tardos, KDD’03 Independent Cascade Model o Information diffuses over time o Each neighbor who converts has a one-time chance to convert others Linear Threshold Model o Each node considers the preferences of all neighbors o If total weight passes threshold, a node converts

19. Adaptive Hypermedia 2008 19 Video suggestion and discovery for YouTube: Taking random walks through the view graph Shumeet Baluja, et al., Google, WWW 2008

20. Adaptive Hypermedia 2008 20 Research Questions How do preferences propagate naturally? What predicts fads? How do recommenders influence propagation?

21. U NIVERSITY OF M INNESOTA 4. Ebay Online Auctions Customers Selling to Customers

22. Adaptive Hypermedia 2008 22 Google Trends Front Page

23. Adaptive Hypermedia 2008 23

24. Adaptive Hypermedia 2008 24 4Chan vs. eBaumsWorld 4Chan o Google Trends Hack o Chocolate Rain eBaumsWorld o Many other hacks o “copyright” fight with 4chan

25. U NIVERSITY OF M INNESOTA

26. The Internet is Serious Business “A phrase used to remind those who voluntarily leave the house that being mocked on the Internet is, in fact, the end of the world.” - Encyclopedia Dramatica

27. Adaptive Hypermedia 2008 27 Amazon Robertson shilled

28. Adaptive Hypermedia 2008 28 The Information Cost of Manipulation- Resistance in Recommender Systems Resnick and Sami. ACM RecSys 08. The Social Cost of Cheap Pseudonyms Friedman and Resnick, Journal of Economics and Management Strategy, 2001

29. U NIVERSITY OF M INNESOTA Increasing Contributions

30. Adaptive Hypermedia 2008 30 What Theory Tells Us… Collective Effort Model o People will contribute more if:  They believe their effort is important to the group  They like the group Smaller is Better o Slovic, Fischhoff, & Lichtenstein, 1980 o People feel greater concern when the reference group they’re part of grows smaller. Specificity Matters o Small & Loewenstein, 2003 o Specific identity of those helped is important in drawing people’s support.

31. Adaptive Hypermedia 2008 31 CommunityLab Research Social science to increase contributions o Accessible to designers o Algorithms, interfaces, toolkits GroupLens @ Minnesota o Recommender algorithms and interfaces o John Riedl, Joe Konstan, Loren Terveen Bob Kraut and Sara Kiesler @ CMU o Social psychology of computer use Paul Resnick and Yan Chen @ Michigan

32. Adaptive Hypermedia 2008 32 VOICE 2 Screen shot Numerical values are represented by smilies Who the contribution helps Value of each contribution

33. Adaptive Hypermedia 2008 33 Results Want Smilies on the regular interface? Self-report Self 3.87 All MovieLens 3.13 Similar Group 2.97 Dissimilar Group 2.94 Control 2.68 Probability of rating a movie Behavioral data Self 7.2% All MovieLens 10.2% Similar Group 15.8% Dissimilar Group 5.9% Control 7.4%

34. Adaptive Hypermedia 2008 34 Research Questions How can contributors be motivated? How can social attacks be mitigated? o Mail list “unsubscribe” How does social psychology interact with defense algorithms? o Can the griefers be encouraged to give up? Can freedoms be preserved?

35. U NIVERSITY OF M INNESOTA 5. Yahoo! Everything

36. Adaptive Hypermedia 2008 36 Flickr Popular Tags

37. Adaptive Hypermedia 2008 37 Tag Selection Algorithms “The Quest for Quality Tags” S. Sen, F. Harper, A. LaPitz, J. Riedl GROUP 2007

38. Adaptive Hypermedia 2008 38 Catcher in the Rye Huge number of tags RQ: How can a tagging system show users tags they want to see?

39. Adaptive Hypermedia 2008 39 Users don’t agree Most controversial tags (Bayesian expected entropy): tagentropy # # comedy0.9872830 classic0.9862524 stylized0.9832021 nudity (full frontal)0.9801820 romance0.9801817 quirky0.9772520 magic0.9741815 animation0.9742620 Steven Spielberg0.97312 sci-fi0.9721417

40. Adaptive Hypermedia 2008 40 Tag Prediction Random baseline: 21% Implicit features: number of applications (39%) number of users (51%) number of searches for a tag (44%) number of users who searched for a tag (48%) length of tag (42%) Moderation-based features: global average rating for a tag (59%) user-normalized global average rating for a tag (62%) tag reputation (57%) Hybrid combinations: logistic regression, decision trees (67%)

41. Adaptive Hypermedia 2008 41 Research Questions How can a system distinguish between “good” tags and “bad” tags? How should quality control work? Can folksonomy be encouraged? o Showing users more tags leads to more vocabulary reuse o How much convergence is valuable?

42. U NIVERSITY OF M INNESOTA 6. Wikipedia Next slide, please!

43. Adaptive Hypermedia 2008 43 Wikipedia on Wikipedia

44. U NIVERSITY OF M INNESOTA Wikiality on MySpace 1:20 – 2:15: edit wikipedia to make truth “What if the number of elephants in Africa were increasing?”

45. U NIVERSITY OF M INNESOTA Creating, Destroying, and Restoring Value in Wikipedia Group 2007 Reid Priedhorsky Jilin Chen Shyong (Tony) K. Lam Katherine Panciera Loren Terveen John Riedl

46. Adaptive Hypermedia 2008 46

47. Adaptive Hypermedia 2008 47

48. Adaptive Hypermedia 2008 48

49. Adaptive Hypermedia 2008 49 Who contributes Wikipedia’s value? User:Maveric149 3.8 million least frequent editors 0.5% of value14% of value Wales Swartz

50. Adaptive Hypermedia 2008 50 PWV contributions of elite editors

51. Adaptive Hypermedia 2008 51

52. Adaptive Hypermedia 2008 52

53. Adaptive Hypermedia 2008 53 Research Questions How can vandalism be detected? How efficient is Wikipedia? How much conflict is valuable?

54. U NIVERSITY OF M INNESOTA 7. Studiverzeichnis Social Network

55. Adaptive Hypermedia 2008 55

56. Adaptive Hypermedia 2008 56

57. Adaptive Hypermedia 2008 57

58. Adaptive Hypermedia 2008 58 The Predictive Power of Online Chatter Gruhl, Guha, Kumar, Novak, Tomkins Yahoo ACM KDD 2005 Volume of blog postings predict sales rank of books Queries can be automatically generated in many cases. Can sometimes predict spikes in sales rank.

59. Adaptive Hypermedia 2008 59 Anti-aliasing on the Web Jasmine Novak, Prabhakar Raghavan, Andrew Tomkins. WWW 2004

60. Adaptive Hypermedia 2008 60 Zip Birthdate Sex Story: Finding Medical Records (Sweeney 2002) Medical Data Ethnicity Visit Date Diagnosis Procedure Medication Total Charge Voter List Name Address Date registered Party affiliation Date last voted Zip Birthdate Sex Former Governer of Massachussetts!

61. Adaptive Hypermedia 2008 61 Risk of Information Exposure (Frankowski et al., SIGIR ‘06) Sparse Dataset 1: private YOU Sparse Dataset 2: public YOU + + = Your private data revealed! Combining algs Keep private information within domain!

62. Adaptive Hypermedia 2008 62 MovieLens Forums -Started June 2005 -Users talk about movies -Public: on the web, no login to read -Can people identify these users in our anonymized dataset?

63. Adaptive Hypermedia 2008 63 Research Questions Can users be identified from the personal recommendation data? YES Can the datasets be redacted to protect the users? UNKNOWN Can the users be warned in time? OPEN QUESTION

64. Adaptive Hypermedia 2008 64

65. Adaptive Hypermedia 2008 65

66. Adaptive Hypermedia 2008 66 Quantity Quality Tags Social Identity ResearchPractice ConceptUnderstanding

67. Adaptive Hypermedia 2008 67 Messages Community-maintained Artifacts of Lasting Value o Requires User Modeling and Adaptive Hypermedia Key Research Challenges: o Attract contributions o Maintain quality o Achieve agreement

68. Adaptive Hypermedia 2008 68 Acknowledgements GroupLens o John Riedl, Joe Konstan, Loren Terveen o Dan Cosley, Shilad Sen, Tony Lam, Rich Davies, Dan Frankowski, Max Harper, Sara Drenner, Al Mamunur Rashid, Sean McNee, Reid Priedhorsky, Aaron Halfaker CommunityLab o Sara Kiesler, Bob Kraut, Paul Resnick, Yan Chen NSF o DGE 95-54517, IIS 96-13960, IIS 97-34442, IIS 99-78717, IIS 01-02229, IIS 03- 24851, IIS 05-34420, IIS 03-25837