Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Information Seeking with Social Signals: Anatomy of aSocial Tag-based Exploratory Search Browser Ed H. Chi, Rowan Nairn Pa...
Social Search Survey [Evans & Chi, CSCW2008] <ul><li>150 user surveys </li></ul><ul><li>Help understand the importance of:...
TagSearch Exploratory Focus 3 kinds of search navigational transactional 28% 13% You know what you want and where it is Yo...
Research Motivation
 
MapReduce Implementation <ul><li>Spreading Activation in a bigraph </li></ul><ul><li>MapReduce computation over a large da...
Use Semantic Analysis to Reduce Noise Guide Web Howto Tips Help Tools Tip Tricks Tutorial Tutorials Reference Semantic Sim...
TagSearch Architecture <ul><li>MapReduce: months of computation to a single day </li></ul><ul><li>Development of novel sco...
Baseline Interface
Exploratory Interface
Experiment Design <ul><li>2 interface x 3 task domain design </li></ul><ul><ul><li>2 Interface (between-subjects) </li></u...
Page Collection Tasks [6min each]
Summarization Tasks [12min each]
Procedure [2 hours] <ul><li>Prior Knowledge Test </li></ul><ul><li>1 st  Task Domain  </li></ul><ul><ul><li>With easy and ...
Results: Interaction Behaviors <ul><li>Number of Queries </li></ul><ul><ul><li>Effect of Interface on number of queries (p...
Results: Page Collection Task <ul><ul><li>Effects of Task Domain (p<.01) and Task Difficulty (p<.05) </li></ul></ul><ul><u...
Results: Summarization Tasks <ul><ul><li>Quality of summarization scored (Cohen’s Kappa=0.7) </li></ul></ul><ul><ul><li>AN...
Results: Keyword Generation Tasks <ul><ul><li>ANCOVA showed Exploratory > Baseline for  Future Architecture  (p<.05) and  ...
Results: Cognitive Load <ul><ul><li>Exploratory > Baseline (p<.05) </li></ul></ul>
Discussion <ul><li>Exploratory interface users: </li></ul><ul><ul><li>performed more queries,  </li></ul></ul><ul><ul><li>...
Summary <ul><li>Harnessing user-generated tags to enrich content for social search </li></ul><ul><li>Weaknesses of social ...
Thanks! <ul><ul><li>http://mrtaggy.com </li></ul></ul><ul><ul><li>http://spartag.us </li></ul></ul><ul><ul><li>http://wiki...
Research Vision Augmented Social Cognition <ul><li>Cognition : the ability to remember, think, and reason; the faculty of ...
Augmented Social Cognition Higher Productivity via Collective Intelligence Intelligence that emerges from the collaboratio...
You’ve finished this document.
Download and read it offline.
Upcoming SlideShare
Google Tag Manager - 1 Tag To Rule Them All (GABC 2013)
Next
Upcoming SlideShare
Google Tag Manager - 1 Tag To Rule Them All (GABC 2013)
Next
Download to read offline and view in fullscreen.

Share

Information Seeking with Social Signals: Anatomy of a Social Tag-based Exploratory Search Browser

Download to read offline

Presented at IUI2010 conference workshop on Social Recommender Systems

http://mrtaggy.com

Related Books

Free with a 30 day trial from Scribd

See all

Related Audiobooks

Free with a 30 day trial from Scribd

See all

Information Seeking with Social Signals: Anatomy of a Social Tag-based Exploratory Search Browser

  1. 1. Information Seeking with Social Signals: Anatomy of aSocial Tag-based Exploratory Search Browser Ed H. Chi, Rowan Nairn Palo Alto Research Center Contact: [email_address] Area Manager, Augmented Social Cognition Area
  2. 2. Social Search Survey [Evans & Chi, CSCW2008] <ul><li>150 user surveys </li></ul><ul><li>Help understand the importance of: </li></ul><ul><ul><li>social cues and information exchanges </li></ul></ul><ul><ul><li>vocabulary problems </li></ul></ul><ul><ul><li>distribution and organization </li></ul></ul>
  3. 3. TagSearch Exploratory Focus 3 kinds of search navigational transactional 28% 13% You know what you want and where it is You know what you want to do Existing search engines are OK informational 59% You roughly know what you want but don’t know how to find it Difficult for existing search engines Opportunity
  4. 4. Research Motivation
  5. 6. MapReduce Implementation <ul><li>Spreading Activation in a bigraph </li></ul><ul><li>MapReduce computation over a large data set </li></ul><ul><ul><li>150 Million+ bookmarks </li></ul></ul>Tags URLs P(URL|Tag) P(Tag|URL)
  6. 7. Use Semantic Analysis to Reduce Noise Guide Web Howto Tips Help Tools Tip Tricks Tutorial Tutorials Reference Semantic Similarity Graph
  7. 8. TagSearch Architecture <ul><li>MapReduce: months of computation to a single day </li></ul><ul><li>Development of novel scoring function </li></ul>
  8. 9. Baseline Interface
  9. 10. Exploratory Interface
  10. 11. Experiment Design <ul><li>2 interface x 3 task domain design </li></ul><ul><ul><li>2 Interface (between-subjects) </li></ul></ul><ul><ul><ul><li>Exploratory vs. Baseline </li></ul></ul></ul><ul><ul><li>3 task domains (within-subjects) </li></ul></ul><ul><ul><ul><li>Future Architecture, Global Warming, Web Mashups </li></ul></ul></ul><ul><li>30 Subjects (22 male, 8 female) </li></ul><ul><ul><li>Intermediate or advanced computer and web search skills </li></ul></ul><ul><ul><li>Half assigned Exploratory, half Baseline. </li></ul></ul><ul><li>For each domain, single block with 3 task types: </li></ul><ul><ul><li>Easy and Difficult Page Collection Task [6min each] </li></ul></ul><ul><ul><li>Summarization Task [12min] </li></ul></ul><ul><ul><li>Keyword Generation Task [2min] </li></ul></ul>
  11. 12. Page Collection Tasks [6min each]
  12. 13. Summarization Tasks [12min each]
  13. 14. Procedure [2 hours] <ul><li>Prior Knowledge Test </li></ul><ul><li>1 st Task Domain </li></ul><ul><ul><li>With easy and difficult page collection tasks, summarization and keyword generation task. </li></ul></ul><ul><ul><li>NASA cognitive load questionnaire </li></ul></ul><ul><li>2 nd Task Domain </li></ul><ul><ul><li>Same battery of tasks and cognitive load questionaire </li></ul></ul><ul><li>3 rd Task Domain </li></ul><ul><li>Experimental Survey </li></ul>
  14. 15. Results: Interaction Behaviors <ul><li>Number of Queries </li></ul><ul><ul><li>Effect of Interface on number of queries (p < .01) </li></ul></ul><ul><ul><ul><li>Exploratory (M=7.81) > Baseline (M=3.77) </li></ul></ul></ul><ul><li>Time Taken </li></ul><ul><ul><li>Effect of Interface on time taken (p < .01) </li></ul></ul><ul><ul><ul><li>Exploratory (7.7min) > Baseline (6.6min) </li></ul></ul></ul>
  15. 16. Results: Page Collection Task <ul><ul><li>Effects of Task Domain (p<.01) and Task Difficulty (p<.05) </li></ul></ul><ul><ul><li>Interaction effect of Interface by Task Domain (p<.05), with Exploratory interface performing better in the Web Mashup domain </li></ul></ul><ul><ul><li>For relevance scores, similar patterns. </li></ul></ul><ul><ul><li>Measure of # of pages collected </li></ul></ul>
  16. 17. Results: Summarization Tasks <ul><ul><li>Quality of summarization scored (Cohen’s Kappa=0.7) </li></ul></ul><ul><ul><li>ANCOVA with Prior Knowledge as covariate </li></ul></ul><ul><ul><li>Exploratory Interface scored higher in Future Architecture (p<.05) and Global Warming (p<.05) </li></ul></ul><ul><ul><li>For Web Mashup, Prior Knowledge correlated positively with performance (r=.51) </li></ul></ul>
  17. 18. Results: Keyword Generation Tasks <ul><ul><li>ANCOVA showed Exploratory > Baseline for Future Architecture (p<.05) and Web Mashups (p<.01), but not for Global Warming. </li></ul></ul><ul><ul><li>Linear model between PK and # of keyword generated for Baseline showed mean slope = 0.32 and significant (p<.05) </li></ul></ul>
  18. 19. Results: Cognitive Load <ul><ul><li>Exploratory > Baseline (p<.05) </li></ul></ul>
  19. 20. Discussion <ul><li>Exploratory interface users: </li></ul><ul><ul><li>performed more queries, </li></ul></ul><ul><ul><li>took more time, </li></ul></ul><ul><ul><li>wrote better summaries (in 2/3 domains), </li></ul></ul><ul><ul><li>generated more relevant keywords (in 2/3 domains), and </li></ul></ul><ul><ul><li>had a higher cognitive load. </li></ul></ul><ul><li>Suggestive of deeper engagement and better learning. </li></ul><ul><li>Some evidence of scaffolding for novices in the keyword generation and summarization tasks. </li></ul>
  20. 21. Summary <ul><li>Harnessing user-generated tags to enrich content for social search </li></ul><ul><li>Weaknesses of social tagging systems is Tag Noise and Inconsistency </li></ul><ul><ul><li>Difficult to leverage for search </li></ul></ul><ul><ul><li>Use data mining techniques to normalize and reduce noise </li></ul></ul><ul><ul><li>Apply normalized tag data in new search algorithm </li></ul></ul><ul><li>Study suggest deeper user engagement in exploration and better learning with MrTaggy </li></ul>
  21. 22. Thanks! <ul><ul><li>http://mrtaggy.com </li></ul></ul><ul><ul><li>http://spartag.us </li></ul></ul><ul><ul><li>http://wikidashboard.parc.com </li></ul></ul><ul><ul><li>Our Blog: http://asc-parc.blogspot.com </li></ul></ul><ul><ul><li>Contact: </li></ul></ul><ul><ul><li>Ed H. Chi, Ph.D. </li></ul></ul><ul><ul><li>Manager, Augmented Social Cognition Area </li></ul></ul><ul><ul><li>[email_address] </li></ul></ul><ul><ul><li>Kammerer, Y., Nairn, R., Pirolli, P., and Chi, E. H. 2009. Signpost from the masses: learning effects in an exploratory social tag search browser. In Proceedings of the 27th international Conference on Human Factors in Computing Systems (Boston, MA, USA, April 04 - 09, 2009). CHI '09. ACM, New York, NY, 625-634. </li></ul></ul>
  22. 23. Research Vision Augmented Social Cognition <ul><li>Cognition : the ability to remember, think, and reason; the faculty of knowing. </li></ul><ul><li>Social Cognition : the ability of a group to remember, think, and reason; the construction of knowledge structures by a group. </li></ul><ul><ul><li>(not quite the same as in the branch of psychology that studies the cognitive processes involved in social interaction, though included) </li></ul></ul><ul><li>Augmented Social Cognition : Supported by systems, the enhancement of the ability of a group to remember, think, and reason; the system-supported construction of knowledge structures by a group. </li></ul><ul><li>Citation: Ed H. Chi. The Social Web: Opportunities for Research. IEEE Computer, Sept 2008 </li></ul>2008-11-07 Ed H. Chi ASC Overview
  23. 24. Augmented Social Cognition Higher Productivity via Collective Intelligence Intelligence that emerges from the collaboration and competition of many individuals <ul><li>Foundation: </li></ul><ul><li>Understanding of human cognition and behavior </li></ul><ul><li>Data mining of social data </li></ul><ul><li>Generic benefits: </li></ul><ul><li>Greater trust </li></ul><ul><li>Better decision-making </li></ul><ul><li>Useful sharing of info </li></ul><ul><li>Auto-organization thru social data </li></ul>Collective Intelligence search sharing foraging <ul><li>TagSearch: Mining social data for automatic data clustering and organization: </li></ul><ul><ul><li>Better organization via user-assigned tags </li></ul></ul><ul><ul><li>Better UI for browsing interesting contents </li></ul></ul><ul><ul><li>Recommendation instead of just search </li></ul></ul><ul><li>Social Transparency create trust and attribution: </li></ul><ul><ul><li>Increase participation via attribution </li></ul></ul><ul><ul><li>Increase credibility and trust with community feedback </li></ul></ul><ul><ul><li>Reduce wiki risks </li></ul></ul><ul><li>SparTag.us: sharing of interesting contents: </li></ul><ul><ul><li>A notebook that automatically organizes your reading </li></ul></ul><ul><ul><li>Social sharing of important and interesting tidbits </li></ul></ul><ul><ul><li>Viral sharing of highlighted and tagged paragraphs </li></ul></ul>
  • DanF

    Apr. 8, 2016
  • qhero74

    Feb. 25, 2010

Presented at IUI2010 conference workshop on Social Recommender Systems http://mrtaggy.com

Views

Total views

2,041

On Slideshare

0

From embeds

0

Number of embeds

9

Actions

Downloads

45

Shares

0

Comments

0

Likes

2

×