Measuring Semantic Similarity between Words Using Web Search Engines Danushka Bollegala, Yutaka Matsuo, Mitsuru Ishizuka

1
Measuring Semantic Similarity between Words
Using Web Search EnginesDanushka Bollegala,
Yutaka Matsuo, Mitsuru Ishizuka

• Topic
• Semantic similarity measures between two words
• Why interesting?
• In information retrieval
• Query expansion
• Automatic annotation of Web pages
• Community mining
• In natural language processing
• Word-sense disambiguation
• Synonym extraction
• Language modeling

WWW 2007 Paper Presentation Zheshen Wang
May 8th, 2007
2

• Solution proposed
• By using the information available on the Web
• Page Counts Text Snippets
• SVM for an optimal combination

• Page Counts
• Co-occurrence measures Jaccard, Overlap
  (Simpson), Dice, PMI
• Modification Suppress random co-occurrences
• Score0, if H(PnQ)ltc, H(x) page counts for the
  query x

• Text Snippets (context and statistical based)?
  top 200 Pattern Freq
• Lexico -syntactic Patterns Extraction
• e.g. Toyota and Nissan are two major Japanese
  car manufactures.
• If the appearing times of a pattern words
• in snippets for synonymous words gtgt in
  snippets for non-synonymous
• it is a reliable indicator of synonymy.

• Combination
• 204-D Feature vector F 200 Pattern Freq, 4
  co-occurrence measures
• Two-class SVM
• synonymous word-pairs (Positive), non-synonymous
  word-pairs (Negative)

WWW 2007 Paper Presentation Zheshen Wang
May 8th, 2007
3
My criticisms of the solution

• Statistics and context based pattern selection is
  not reliable (No ontology or syntax templates)
• Sparse Distribution
• Noises (meaningless patterns)
• Correlations (e.g. X and Y , X and Y are, X
  and Y are two)
• Missing meaningful patterns due to limited
  n-grams range
• (e.g. X and Y are far apart, beyond the range of
  n-grams, n2,3,4,5
• Rose is a very popular flower in
  the US.)
• Feature vector F 200 Pattern Freq, 4
  co-occurrence measures
• Error prone for uncommon words
• e.g. rarely used professional terms
• Base set from the web is too small to be
  reliable.
• Like the case of CBioC, users voting would be
  better

WWW 2007 Paper Presentation Zheshen Wang
May 8th, 2007
4
How it is related to our course?

• Web-based information extraction (Knowledge
  Extraction)
• Extract base level knowledge (facts) directly
  from the web
• Page counts(Hits), e.g. Knowitall
• Inevitable drawback Error prone for uncommon
  words in the web, e.g. CBioC
• Making use of Collective UnconsciousBig Idea 3
• Analyzing term co-occurrences to capture semantic
  information
• Co-occurrence measures
• Similarity measure in terms of co-occurrence
• Jaccard, Overlap (Simpson), PMI
• Making use of context based on statistics
• Patterns from context rather than from an
  ontology (SemTag Seeker).
• Patterns decided by statistics rather than
  templates from syntax tree (Generic extraction
  patterns, Hearst 92).
• n-grams for a word, somewhat like the
  20-word-window of spot(l,c) in SemTag
  Seeker.

WWW 2007 Paper Presentation Zheshen Wang
May 8th, 2007
5
Measuring Semantic Similarity between Words Using
Web Search EnginesDanushka Bollegala, Yutaka
Matsuo, Mitsuru Ishizuka
WWW 2007 Paper Presentation Zheshen Wang
May 8th, 2007