Social Awareness Streams (SAS) <ul><li>Short, natural-language messages created by users </li></ul><ul><li>Broadcasted </l...
31 May 2010 What‘s the knowledge of SAS?  Users Web Resources Real-World Happenings Natural Language Constructs http://www...
Proposed Approach  <ul><li>Aim </li></ul><ul><ul><li>What kind of knowledge is contained in SAS </li></ul></ul><ul><ul><li...
Research Questions <ul><li>Do ontological structures emerge from SAS?  </li></ul><ul><li>Which factors influence their emer...
SAS Analyzer 31 May 2010
Controlled Experiments (1) 31 May 2010 Do ontological structures emerge from SAS? ground truth ontology randomly sample  I...
Controlled Experiments (2) 31 May 2010 Which factors influence emerging semantics? vary variables Compare (e.g., via Precis...
Preliminary Results <ul><li>Network-theoretic Model of SAS </li></ul><ul><li>Structural Stream Measures </li></ul><ul><li>...
A network-theoretic model of SAS <ul><li>A  Social Awareness Stream  is a tupel </li></ul><ul><li>U, M and R are finite se...
Example 31 May 2010
Structural Stream Measures (1) 31 May 2010
Structural Stream Measures (2)  <ul><li>Social Diversity </li></ul><ul><ul><li>How many different users participate in a s...
Experiment <ul><li>Aim </li></ul><ul><ul><li>Can we observe emerging semantics from SAS? </li></ul></ul><ul><li>Method </l...
Dataset <ul><li>4 different stream aggregations from Twitter </li></ul><ul><li>Same topic </li></ul><ul><ul><li>Hashtag st...
Network Transformations 31 May 2010 co-occurence context [ Harris , 1954] [Mika, 2007] communities
First Results (1) <ul><li>Type of stream aggregations influence emerging semantics </li></ul><ul><ul><li>Hashtag stream ag...
First Results (2) <ul><li>Type of network transformation influence emerging semantics  </li></ul><ul><ul><li>Hashtags seem...
Limitations <ul><li>Small Dataset </li></ul><ul><li>Only one topic/domain </li></ul><ul><li>Manual Evaluation </li></ul>31...
References <ul><li>Z. Harris. Distributional structure. The Structure of Language: Readings in the philosophy of language,...
Thank you! 31 May 2010 http://clauwa.info/me [email_address] http://twitter.com/clauwa
Upcoming SlideShare
Loading in...5
×

Knowledge Acquisition from Social Awareness Streams

2,769

Published on

Slides from PhD symposiumat ESWC2010 (http://www.eswc2010.org/program-menu/phd-symposium)

Published in: Technology, Education
0 Comments
3 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total Views
2,769
On Slideshare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
54
Comments
0
Likes
3
Embeds 0
No embeds

No notes for slide

Knowledge Acquisition from Social Awareness Streams

  1. 2. Social Awareness Streams (SAS) <ul><li>Short, natural-language messages created by users </li></ul><ul><li>Broadcasted </li></ul><ul><li>Information consumption is driven by social networks </li></ul><ul><li>Applications such as Twitter or Facebook </li></ul><ul><li>[Naaman, 2010] </li></ul>31 May 2010
  2. 3. 31 May 2010 What‘s the knowledge of SAS? Users Web Resources Real-World Happenings Natural Language Constructs http://www.flickr.com/photos/matthewfield/2306001896/ What‘s the knowledge of SAS? Users Web Resources Natural Language Constructs Real-World Happenings http://www.flickr.com/photos/waldoj/722508166/
  3. 4. Proposed Approach <ul><li>Aim </li></ul><ul><ul><li>What kind of knowledge is contained in SAS </li></ul></ul><ul><ul><li>How we can acquire knowledge from SAS </li></ul></ul><ul><ul><li>Which factors influence knowledge acquisition results </li></ul></ul><ul><li>Method </li></ul><ul><ul><li>Develop a SAS Analyzer system </li></ul></ul><ul><ul><li>Controlled experiments </li></ul></ul>31 May 2010
  4. 5. Research Questions <ul><li>Do ontological structures emerge from SAS? </li></ul><ul><li>Which factors influence their emergence? </li></ul><ul><ul><li>Stream aggregation/sampling strategies? </li></ul></ul><ul><ul><li>Semantic enrichment strategies? </li></ul></ul><ul><ul><li>Knowledge acquisition methods and algorithms? </li></ul></ul>31 May 2010
  5. 6. SAS Analyzer 31 May 2010
  6. 7. Controlled Experiments (1) 31 May 2010 Do ontological structures emerge from SAS? ground truth ontology randomly sample Input labels Compare (e.g., via Precision, Recall, RLA [Maedche et al, 2000] )
  7. 8. Controlled Experiments (2) 31 May 2010 Which factors influence emerging semantics? vary variables Compare (e.g., via Precision, Recall, RLA [Maedche et al, 2000] ) ground truth ontology randomly sample Input labels
  8. 9. Preliminary Results <ul><li>Network-theoretic Model of SAS </li></ul><ul><li>Structural Stream Measures </li></ul><ul><li>First Experiment on acquiring latent conceptual structures from SAS </li></ul>31 May 2010
  9. 10. A network-theoretic model of SAS <ul><li>A Social Awareness Stream is a tupel </li></ul><ul><li>U, M and R are finite sets whose elements are called users, messages and resources </li></ul><ul><li>q1, q2, q3 are qualifiers </li></ul><ul><li>Y is a ternary relation </li></ul><ul><li>ft is a function </li></ul><ul><li>fl is a function </li></ul>31 May 2010
  10. 11. Example 31 May 2010
  11. 12. Structural Stream Measures (1) 31 May 2010
  12. 13. Structural Stream Measures (2) <ul><li>Social Diversity </li></ul><ul><ul><li>How many different users participate in a stream? </li></ul></ul><ul><ul><li>Social variety: </li></ul></ul><ul><ul><li>How balanced are their participations? </li></ul></ul><ul><ul><li>Social balance: </li></ul></ul>31 May 2010
  13. 14. Experiment <ul><li>Aim </li></ul><ul><ul><li>Can we observe emerging semantics from SAS? </li></ul></ul><ul><li>Method </li></ul><ul><ul><li>Input: topic of interest, in our case „semanticweb“ </li></ul></ul><ul><ul><li>4 different stream aggregations </li></ul></ul><ul><ul><li>3-mode networks (users, resources and messages) </li></ul></ul><ul><ul><li>Network transformations (projections) to obtain lower-order networks of resources </li></ul></ul><ul><ul><li>Output: weighted resource networks </li></ul></ul><ul><ul><li>Manual evaluation </li></ul></ul>31 May 2010
  14. 15. Dataset <ul><li>4 different stream aggregations from Twitter </li></ul><ul><li>Same topic </li></ul><ul><ul><li>Hashtag stream: #semanticweb </li></ul></ul><ul><ul><li>Keyword stream: semanticweb and semweb </li></ul></ul><ul><ul><li>User list stream: semweb user list from twitter user sclopit </li></ul></ul><ul><ul><li>User directory stream: wefollow semanticweb directory </li></ul></ul><ul><li>Same time interval </li></ul><ul><ul><li>2 time intervals: 16th of Dec 2009 - 20th of Dec 2009 and 29th of Dec 2009 - 1st of Jan 2010 </li></ul></ul>31 May 2010
  15. 16. Network Transformations 31 May 2010 co-occurence context [ Harris , 1954] [Mika, 2007] communities
  16. 17. First Results (1) <ul><li>Type of stream aggregations influence emerging semantics </li></ul><ul><ul><li>Hashtag stream aggregations are more robust against external disturbances than user list streams </li></ul></ul>31 May 2010 Hashtag Stream O R (RU a )S(R h ) User List Stream O R (RU a )S(R UL )
  17. 18. First Results (2) <ul><li>Type of network transformation influence emerging semantics </li></ul><ul><ul><li>Hashtags seem to be good context indicators </li></ul></ul><ul><ul><li>Resource-hashtag networks reveal good latent conceptual structures </li></ul></ul>31 May 2010
  18. 19. Limitations <ul><li>Small Dataset </li></ul><ul><li>Only one topic/domain </li></ul><ul><li>Manual Evaluation </li></ul>31 May 2010
  19. 20. References <ul><li>Z. Harris. Distributional structure. The Structure of Language: Readings in the philosophy of language,10:146-162, 1954. </li></ul><ul><li>A. Maedche and S. Staab. Discovering Conceptual Relations from Text. In: W.Horn (ed.): ECAI 2000. In Proceedings of the 14th European Conference on Artificial Intelligence, Berlin, Amsterdam, 2000. </li></ul><ul><li>P. Mika. Ontologies are us: A unified model of social networks and semantics. Web Semantics, 5(1):5-15, 2007. </li></ul><ul><li>M. Naaman, J. Boase, and C.-H. Lai. Is it all about me? user content in social awareness streams. In Proceedings of the ACM 2010 conference on Computer supported cooperative work, 2010. </li></ul>31 May 2010
  20. 21. Thank you! 31 May 2010 http://clauwa.info/me [email_address] http://twitter.com/clauwa
  1. A particular slide catching your eye?

    Clipping is a handy way to collect important slides you want to go back to later.

×