Fw: [Corpora-List] computing semantic word similarity

From: Olga Mitrofanova (alkonost@OM12520.spb.edu)
Date: Wed Nov 16 2005 - 17:58:57 MET

  • Next message: Martin Wynne: "Re: [Corpora-List] free tagged corpus"

    ----- Original Message -----
    From: Olga Mitrofanova
    To: Dimitar Blagoev
    Sent: Wednesday, November 16, 2005 7:54 PM
    Subject: Re: [Corpora-List] computing semantic word similarity

    Among recent papers on measuring word-similarity one should mention a book by Dominic Widdows "Geometry and Meaning"
    http://infomap.stanford.edu/book/
    (you can find references and demos for each chapter).

    Certain strategies of measuring semantic distances were developed and discussed with regard to Slavic languages (cf. books by Ju. Apresjan, A. Novikov & E. Jaroslavtseva...). As far as I can judge from my own experience, the most profitable and linguistically adequate strategy of computing semantic similarity of words should combine lexicon-based and corpus-based approaches (like LSA).

    Yours sincerely,
    Olga Mitrofanova



    This archive was generated by hypermail 2b29 : Thu Nov 17 2005 - 10:16:38 MET