![E
goal
effectively retrieve social media content
for known events from multiple services
[with Hila Becker, Luis Gravano]](https://image.slidesharecdn.com/ecirkeynote-130325075605-phpapp01/85/ECIR-2013-Keynote-Time-for-Events-22-320.jpg)
![E
E
for example...
120"
100"
80"
60"
40"
20"
0"
6/7/11" 6/8/11" 6/9/11" 6/10/11" 6/11/11" 6/12/11" 6/13/11"
[andrew"bird"concert]" [state"farm"insurance]"](https://image.slidesharecdn.com/ecirkeynote-130325075605-phpapp01/85/ECIR-2013-Keynote-Time-for-Events-30-320.jpg)
![research questions
can Twitter content around broadcast
news events inform journalistic inquiry?
what insights and analyses can we
enable through visual analytic tools?
[with postdoctoral fellow Nick Diakopoulos]](https://image.slidesharecdn.com/ecirkeynote-130325075605-phpapp01/85/ECIR-2013-Keynote-Time-for-Events-35-320.jpg)
![what the hell?
[with: Lyndon Kennedy, Dan Ellis, Kai Su]](https://image.slidesharecdn.com/ecirkeynote-130325075605-phpapp01/85/ECIR-2013-Keynote-Time-for-Events-42-320.jpg)
ECIR 2013 Keynote - Time for Events
- 1. time for events telling the world’s stories from social media Mor Naaman Rutgers SC&I & Mahaya, Inc. @informor
- 5. (JCDL 2007)
- 6. (JCDL 2007)
- 7. (SIGIR 2007) yes.
- 8. organize the world’s memories
- 10. BYOBW
- 15. organize the world’s memories
- 16. detect identify organize objectives
- 17. objectives detect ICWSM 2011a JASIST 2011 WebDB 2009 SIGIR 2007
- 18. objectives identify WSDM 2012 ICWSM 2011b WSDM 2010
- 19. objectives organize ICMR 2012 CHI 2012 CSCW 2012 MTAP 2012 VAST 2010 WWW 2009
- 20. today organize identify detect Vox! multi-site Multiplayer
- 21. overview E Multi-site content Vox Civitas Multiplayer
- 22. E goal effectively retrieve social media content for known events from multiple services [with Hila Becker, Luis Gravano]
- 23. E
- 24. E challenges event descriptor not well-formed brief textual descriptors noise formats/conventions/metadata differ
- 25. E approach two-step query formulation precision-based recall-based validate queries based on known/ extracted event model
- 26. E E step 1 term extraction from event descriptors generates “high precision” queries e. g. “andrew bird, opening gala, celebrate brooklyn, prospect park”
- 27. E E step 2 use “high precision” corpus to generate more general queries to improve recall e. g. “andrew bird concert”, “state farm insurance”
- 28. E E recall-oriented queries Benefits: - Works cross-site - Works with short content Challenges: - Introduces noise - Potentially large set of queries
- 29. E E post-filtering use known event model (topics, time, location) use queries with a result set that matches known model
- 30. E E for example... 120" 100" 80" 60" 40" 20" 0" 6/7/11" 6/8/11" 6/9/11" 6/10/11" 6/11/11" 6/12/11" 6/13/11" [andrew"bird"concert]" [state"farm"insurance]"
- 31. E evaluation 1.1" query generation4" 1" 4" 0.9" 0.8" 5" 5" Precision" 0.7" relevance of36"retrieved documents NDCG% 0.6" 39" 34" 34" Twi7er8MS" 0.5" 0.4" 0.3" YouTube8MS" 7" 0.2" 9" 8" 8" 0.1" 0" 0" 5" 10" 15" 20" 25" Number%of%Documents%k%
- 32. E takeaways can aggregate content fragmented across platforms improve recall, not rely on site-specific features
- 33. overview E Multi-site content (WSDM 2012) Vox Civitas Multiplayer
- 35. research questions can Twitter content around broadcast news events inform journalistic inquiry? what insights and analyses can we enable through visual analytic tools? [with postdoctoral fellow Nick Diakopoulos]
- 36. supporting analysis direct attention to relevant information automatic content analysis for filtering – relevance – uniqueness / novelty – sentiment – keyword extraction
- 38. how to evaluate? directly evaluate the output of the algorithms (quantitative) deep, extensive evaluation of users’ interaction with the system (qualitative) read more: Olsen (UIST ’07) Naaman (MTAP ’12)
- 39. Vox evaluation goals • How effective for generating story ideas? • What kind of insights/analysis are supported? • Shortcomings and how features are used?
- 40. takeaways can extract reliable event structure from social media
- 41. overview E Multi-site content Vox Civitas (VAST 2010) Multiplayer
- 42. what the hell? [with: Lyndon Kennedy, Dan Ellis, Kai Su]
- 47. supporting analysis extract the signal from people’s attention: find overlapping moments compute and rank scenes extract scene descriptors
- 48. audio fingerprinting Wang et al. (ISMIR ’03)
- 49. two clips, aligned 0:18 3:32 0:00 0:00 2:32
- 50. a story of n clips time
- 51. from clips to scenes Higher Ground Encore time Happy Birthday, Birthday
- 53. evaluation quantitative: evaluated matching, scene extraction… qualitative: evaluated deployment scenario/task
- 54. takeaways can create an event presentation that gets better them more content is added
- 55. overview E Multi-site content Vox Civitas Multiplayer (NM&S 2012, ICMR 2012, MTAP 2012, WWW 2009)
- 57. towards better models of large-scale human attention
- 58. printing press
- 62. the web
- 64. social media
- 66. new methods?
- 67. search by subject code?
- 68. explore. new information seeking tasks (and models) new applications for social media content
- 69. explore. beyond real-time personal and social
- 70. questions? mor@rutgers.edu @informor http://mornaaman.com
- 71. thanks Luis Gravano Hila Becker Nick Diakopoulos Kai Su Dan Ellis Munmun de Choudhury Tarikh Korula …