ABSTRACT

Word sense disambiguation (WSD) is an open problem of natural language processing, which governs the process of identifying the appropriate sense of a word (i.e. intended meaning) in a sentence, when the word has multiple meanings. In this paper we introduce a new WordNet database relation structure whose usage enhances the WSD efficiency of knowledge-based contextual overlap dependent WSD algorithms, such as the popular Lesk algorithm. The efficiency of WSD, on the usage of the proposed WordNet over existing WordNet as a knowledge-base, has been experimentally verified by using the Lesk algorithm on a rich collection of heterogeneous sentences. Use of the proposed WordNet for Lesk Algorithm highly increases the chances of contextual overlap, thereby resulting in high accuracy of proper sense or context identification of the words. The WSD results and accuracies, obtained using the proposed WordNet, have been compared with the results obtained using existing WordNet. Experimental results show that use of our proposed WordNet results in better accuracy of WSD than the existing WordNet. Thus its usage will help the users better, in doing Machine translation, which is one of the most difficult problems of natural language processing