Knowledge-based Word Sense Disambiguation with Feature Words Based on Dependency Relation and Syntax Tree

Abstract

Context feature words are important for word sense disambiguation (WSD). There are two kinds of methods to extract feature words: window-based and dependency relation method. Both of them have some defects. In order to solve the problems of the existing methods, this paper proposes a knowledgebased WSD method which obtains context feature words by dependency relation and syntax tree. Firstly, according to dependency relation between context words and the ambiguous word, the layer relation and path distance in the phrase structure syntax tree, direct distance in the sentence, feature words are selected from the context and are assigned different WSD weights. Secondly, Based on WordNet (WN), semantic relatedness between each sense of the ambiguous word and feature words are computed and the sense with most semantic relatedness is selected as the right sense. Evaluation is performed over a publicly available lexical sample dataset. The results show that our WSD method is better than the methods that obtain feature words with window or dependency relation. The method is a preferred strategy to select feature words to disambiguate the target words.