Semantic distance

1
Semantic distance WordNet

• Serge B. Potemkin
• Moscow State University
• Philological faculty

2
Distance and metrics

• Fundamental concept
• distance between entities under consideration
• Semantic distance between words or concepts
• Metrical space axioms?

3
Distance is needed for

• word sense disambiguation,
• determining the structure of texts,
• text summarization and annotation,
• information extraction and retrieval,
• automatic indexing,
• lexical selection,
• the automatic correction of word errors in text

4
Approaches to distance measuring

• Corpora-based
• Dictionary-based
• Roget-structured thesauri
• WordNet and other semantic networks

5
WordNet

• Synonym sets (synsets)
• Subsumption hierarchy (hyponymy / hypernymy),
• 3 meronymic (PART-OF) relations
• COMPONENT-OF,
• MEMBER-OF,
• SUBSTANCE-OF and their inverses
• Antonymy,
• COMPLEMENT-OF

6
WordNet shortcomings

• 150000 synsets inadequate coverage
• Non-English versions 20 70 of English
• (100000 synsets for Russian)
• Extension is hard
• Distance measuring is controversial

7
Corpora-based approach

• Two words wa and wb are as close as often their
  neighbors (/- 5 words) coincide.

• Ex. (distributional profile of the word)
• star space 0.28, movie 0.2, famous 0.13,
• light 0.09, rich 0.04, . .

8
Dictionary-based approach

• Two words wa and wb are as close as often words
  in definitions coincide.

• Ex. walinguistics wbstylistics
• the, study, of, language, in, general, and, of,
  particular, languages, and, their, structure,
  and, grammar, and, history
• the, study, of, style, in, written, or, spoken,
  language.
• 2 words coincide in definitions

9
Bilingual dictionary approach

• Two words wa and wb are as close as often their
  equivalents coincide.
• ?(Wa, Wb) 1/Sni,
• Where
• S is the sum over all coinciding Russian
  equivalents
• and ni is the number of dictionaries where an
  equivalent occurs
• Or ?(Wa, Wb) S nai?nbi /(aR?bR)

10
Multidimensional scaling

• Semantic network is a graph
• nodes -- words
• edges -- links between words via bilingual
  lexicon
• edge ?(Wa, Wb)
• Immersion of graph is possible to N-dimensional
  space
• where Nnumber of words in the lexicon (gt100000)
• Multidimensional scaling for visualization

11
New synonyms
12
1-neighborhood of accolade

• Links between synonyms (black)
• Links between synonyms from the dictionary
  (green)
• 2 isolated clusters.

13
Dominant in acerbity neighborhood

• ascerbity (?????????) excluded
• cluster (bold lines) derived by Markovian process
• asperity (????????) is the centre of the cluster

14
2 dominants for bicycling (wheelcrook)
15
Adjustable parameters

• - space dimension
• - minimal number of dictionaries linking
  synonyms
• - maximal distance from the word under
  consideration
• - maximal number of displayed words
• - word excluded from clustering

16
Compare LDB with WordNet (accolade)
n noun, v - verb
17
Controversy 1

• Immediate hyperonym for the accolade synset in
  WordNet is symbol -- (an arbitrary sign (written
  or printed) that has acquired a conventional
  significance).
• Immediate hyperonym for commendation, (more
  frequent than accolade) is accolade synset
• Actually accolade is hyponym for commendation
• It is impossible to disambiguate accolade
  (bracket) from accolade (praise)

18
Controversy 2

• WordNet
• dog 1 domestic dog
• hyperonym - canine, canid.
• further mammal, , entity
• Nor animal, neither pet, are linked with dog as
  hyperonyms.
• Tree structure is inadequate for semantic coding.

19
Conclusion

• Each meaning of the polysemic word could be coded
  as pair (wE, wR) in contrast to synset coding.
• Metrics superimposed over LDB enables homograph
  disambiguation and extraction of dominants
• Network has particular advantages over
  hierarchical representation of semantic relations