Word sense disambiguation using WordNet and the Lesk algorithm
Jonas Ekedahl, Koraljka Golub · 2004
Word sense disambiguation is the process of automatically clarifying the meaning of a word in its context. It has drawn much interest in the last decade and much improved results are being obtained. In this paper we take the so-called Lesk approach. In our case, definitions of the senses of the words to be disambiguated, as well as of the ten surrounding nouns, adjectives and verbs, are derived and enriched using the WordNet lexical database. Two possible implications of this project could be that the results are dependent on the characteristics of a test document and on the characteristics of glosses, which needs to be further investigated. The average precision performed worse (0.45) than baseline precision (0.60) which was based on always selecting the most frequent sense. However, the presented approach has several limitations: a small sample, and a big number of fine senses in WordNet, many of which are not that distinguishable from each other. The future work would include experimenting with different variations of the approach.