Optimizing Web Search Using Social Annotations

1
Optimizing Web Search Using Social Annotations
Shanghai JiaoTong University IBM China
Research Lab
Stated Problem/Why interesting?
Proposed Solution
SocialSimRank to calculate similarity between
annotations and query
Recent studies

• Ranking web pages based on similarity
  (vector-space model)
• Ranking web pages based on quality (PageRank)
• Utilizing social annotations (folksonomies,
  Semantic Web)

Trivial approach
Better solution
First to propose integrating social annotations
with traditional web search
SocialPageRank to calculate quality in terms of
annotation count
My Opinion

• Similar to PageRank quality is passed between
  the users, annotations and pages until convergence

• Annotations are descriptive, but not clean
  summaries (synonymy problem)
• Not very clear exactly how to incorporate both
  algorithms into web search (not the point of the
  paper?)
• Webs adversarial nature ignored, tag spamming
  only briefly mentioned
• Paper is interesting and readable known concepts
  applied in novel way. We couldve thought of it!
  It was our corn to cut!

Connection with Course Material

• Including new measure into basic similarity
  formula
• SocialSimRank label disambiguation approach
  (similar to SemTag/SemSeeker)
• SocialPageRank concept identical to PageRank
• Use of annotation correlation matrices analogous
  to term-term matrices