INFORMATION RETRIEVAL

1
INFORMATION RETRIEVAL

• -Monday March 21, 2006
• -Arjun Dasgupta

2
IR DOMAINS

• Ranking functions in Traditional IR
• Ranking functions used in the WWW
• Ranking functions used in databases

3
RANKING FUNCTOINS IN TRADITIONAL IR

• Started as a branch of library science
• Definitions
• Corpus a collection of documents
• Document a set/bag of words
• Keyword Query small set of keywords

4
An Example.

• Query over the web
• Microsoft, Corporation
• An obvious approach would be to retrieve all
  documents containing Microsoft and
  Corporation ordered by the ranking function
• Ranking function determines the order in which
  query results are presented

5
What constitutes a good ranking function???

• Relative frequency of occurrence of keywords
• Proximity of keywords (this cannot be done in a
  bag of words model)
• Specificity or importance
• Links to/from other documents
• Popularity/ relevance of page with respect to the
  query