Contextual Word Similarity and Estimation from Sparse Data

Ido Dagan, Shaul Marcus and Shaul Markovitch. Contextual Word Similarity and Estimation from Sparse Data. In Proceedings of the 31st Annual Meeting of the Association for Computational Linguistics, 164-171 Ohio State University, 1993.

Abstract

In recent years there is much interest in word cooccurrence relations, such as n-grams, verb-object combinations, or cooccurrence within a limited context. This paper discusses how to estimate the probability of cooccurrences that do not occur in the training data. We present a method that makes local analogies between each specific unobserved cooccurrence and other cooccurrences that contain similar words, as determined by an appropriate word similarity metric. Our evaluation suggests that this method performs better than existing smoothing methods, and may provide an alternative to class based models.

Keywords: Information Retrieval, Semantic Relatedness

Secondary Keywords:

Online version:

Bibtex entry:

 @inproceedings{Dagan:1993:CWS,
  Author = {Ido Dagan and Shaul Marcus and Shaul Markovitch},
  Title = {Contextual Word Similarity and Estimation from Sparse Data},
  Year = {1993},
  Booktitle = {Proceedings of the 31st Annual Meeting of the Association for Computational Linguistics},
  Pages = {164--171},
  Address = {Ohio State University},
  Url = {http://www.cs.technion.ac.il/~shaulm/papers/pdf/Dagan-Marcus-Markovitch-acl1994.pdf},
  Keywords = {Information Retrieval, Semantic Relatedness},
  Secondary-keywords = {Word Cooccurrence, Word Similarity},
  Abstract = {
    In recent years there is much interest in word cooccurrence
    relations, such as n-grams, verb-object combinations, or
    cooccurrence within a limited context. This paper discusses how to
    estimate the probability of cooccurrences that do not occur in the
    training data. We present a method that makes local analogies
    between each specific unobserved cooccurrence and other
    cooccurrences that contain similar words, as determined by an
    appropriate word similarity metric. Our evaluation suggests that
    this method performs better than existing smoothing methods, and
    may provide an alternative to class based models.
  }

  }