A word sense disambiguation algorithm for information retrieval applications

After a brief introduction and overview of some remarkable studies about Natural Language Processing and Word Sense Disambiguation, the authors describe a general purpose statistic method for the disambiguation of polysemous words in English. Unlike other methods, the one here introduced needs no linguistic or grammatical restrictions to produce effective results. Such a method consists in an algorithm based on the co-occurrence and frequency of words in the Internet, and avails itself of WordNet 1.6 and Altavista Search Engine. The results of this study are also presented and discussed. In the final section of the paper, possible future applications for the devised algorithm are described.