Чираг Шах «Коллективный поиск, взаимодействие пользователей: подходы к изучению социальных аспектов информационного поиска»
2 likes496 views
Dr. Chirag Shah presented on distributed and collaborative aspects of information seeking. He discussed how community question answering sites like Yahoo Answers leverage the wisdom of crowds to quickly provide relevant answers at low cost compared to expert-based question answering. However, expert-based sites may provide higher quality, customized responses. Shah proposed hybrid models that link expert and community question answering or facilitate collaboration between experts. He also discussed systems for mediating collaboration between users during information seeking tasks.
![System mediated collaboration [SIGIR 2008]
Prospector Miner
Prospector – Explores promising Miner – Digs deeper looking
areas for mining spots for relevant information](https://image.slidesharecdn.com/chiragshahyandex2012talk-120821024141-phpapp01/85/-27-320.jpg)
![System-mediated asymmetric [IP&M 2010]
Gatherer Surveyor
Gatherer– Explores highly Surveyor – Digs further to
promising information (higher explore more diverse
relevance) information (higher novelty)](https://image.slidesharecdn.com/chiragshahyandex2012talk-120821024141-phpapp01/85/-34-320.jpg)
![Adding a new dimension to information seeking
Applications:
Product search/shopping
Health-related search
Cross-disciplinary projects
Location-based tasks
Online gaming
Synergic solutions
[SIGIR 2011]](https://image.slidesharecdn.com/chiragshahyandex2012talk-120821024141-phpapp01/85/-35-320.jpg)
Чираг Шах «Коллективный поиск, взаимодействие пользователей: подходы к изучению социальных аспектов информационного поиска»
- 1. Distributed Searching, United Searchers: Investigating Social and Collaborative Aspects of Information Seeking Dr. Chirag Shah School of Communication & Information (SC&I) Rutgers, The State University of New Jersey chirags@rutgers.edu http://comminfo.rutgers.edu/~chirags/ @chirag_shah
- 3. Information Seeking Overview of the Talk
- 4. Information behavior Retrieval Searching Answer Seeking Extraction Filtering Synthesis Crawling • Link Information need Harvesting • Summarize Sense-making • Perceive • Recognize • Gist • Understand • Interpret • Analyze • Evaluate
- 5. Information seeking Interaction Reference Query interview Question Collaborative Information Seeking
- 6. Rest of the talk Community Q&A (CQA) Collaborative information seeking (CIS)
- 7. Community Q&A (CQA) Crowdsourcing information seeking
- 8. Question to the crowd
- 9. Online Q&A
- 10. Online Q&A Community Q&A Social Q&A Yahoo! Answers Facebook AnswerBag Twitter AskVille Expert-based Q&A Collaborative Q&A Google Answers WikiAnswers Virtual reference Wikipedia Reference Desk
- 11. Yahoo! Answers
- 12. Seeking the wisdom of crowd
- 13. Seeking the wisdom of crowd
- 14. Our collections Yahoo! Answers Questions: 3 million+ Answers: 16 million+ Avg. answers per question: 5+ Users: 1 million+ Google Answers Questions: 153,343 Answers: 56,835 Avg. answers per question: 0.37 Users: 83,454 + 534
- 15. Google Answers vs. Yahoo! Answers
- 16. Distributions of Users on Yahoo! Answers
- 18. Expert-based vs. Community Q&A Expert User 1. Relevance 1. Speed 2. Accuracy 2. Relevance 3. Authenticity 3. Accuracy 4. Speed 4. Authenticity
- 19. Yahoo! Answers: Time for the First Answer
- 20. Yahoo! Answers: Time for the Best Answer
- 21. Expert-based vs. Community Q&A EQA better than CQA CQA better than EQA • Customization • Cost • Quality • Volume • Relevance • Speed • Accuracy • Social aspects • Authenticity • Engagement • Completeness • Collaboration • Process-focused • Product-focused
- 22. Creating synergic solutions for Q&A Scenario-1: Collaboration among EQA services Solution Challenges • Integration of services • Question routing • Transparent/opaque? EQA1 EQA2 EQA3
- 23. Creating synergic solutions for Q&A Scenario-2: Linking EQA and CQA Solution Challenges • Evaluating content • Personalization • Service policies EQA CQA
- 24. Challenges and opportunities Use the wisdom of the “wise” crowd Social search – making hyper-links to “hyper-connections” Connect Web search with online Q&A for a hybrid model of information seeking
- 25. Collaborative Information Seeking The Art and Science of Making Whole Greater than the Sum of All
- 26. Collaboration in information seeking Requirement or setup e.g., merger Division of labor e.g., class assignment Diversity of skills e.g., co-authorship
- 27. System mediated collaboration [SIGIR 2008] Prospector Miner Prospector – Explores promising Miner – Digs deeper looking areas for mining spots for relevant information
- 28. System mediated collaboration User 1 User 2 User Interface Regulator Input Output (Roles) Coordinator Coordinator Collaborative Algorithmic Search Mediation Engine
- 30. Prospector Miner R E S U L T S Queries
- 31. Prospector Miner
- 32. Prospector Miner
- 33. Miner queue priority Weighted Borda Count fusion prioritydoc Bordadoc,q wseen,q wrel,q q wseen,q N seen,q / N unseen,q wrel,q N rel,q / N nonrel,q
- 34. System-mediated asymmetric [IP&M 2010] Gatherer Surveyor Gatherer– Explores highly Surveyor – Digs further to promising information (higher explore more diverse relevance) information (higher novelty)
- 35. Adding a new dimension to information seeking Applications: Product search/shopping Health-related search Cross-disciplinary projects Location-based tasks Online gaming Synergic solutions [SIGIR 2011]
- 36. User mediated collaboration: Ariadne
- 37. User mediated collaboration: SearchTogether
- 38. User mediated collaboration: Coagmento
- 39. Challenges and opportunities Extending/replacing individual information seeking/retrieval models with those incorporating collaboration Context-dependent analysis of costs and benefits of working in collaboration Integration of collaborative tools into existing practices Converting social tie into collaborative tie and vice versa Evaluation
- 41. Thank you!