Feature Words Selection for Knowledge-based Word Sense Disambiguation with Syntactic Parsing

摘要

Feature words are crucial clues for word sense disambiguation. There are two methods to select feature words: window-based and dependency-based methods. Both of them have some shortcomings, such as irrelevant noise words or paucity of feature words. In order to solve the problems of the existing methods, this paper proposes two methods to select feature words with syntactic parsing, which are based on phrase structure parsing tree (PTree) and dependency parsing tree (DTree). With the help of syntactic parsing, the proposed methods can select feature words more accurately, which can alleviate the effect of noise words of window-based method and can avoid the paucity of feature words of dependency-based method. Evaluation is performed on a knowledge-based WSD system with a publicly available lexical sample dataset. The results show that both of the proposed methods are superior to window-based and dependency-based methods, and the method based on PTree is better than the method based on DTree. Both of them are preferred strategies to select feature words to disambiguate ambiguous words.%W artykule zaproponowano dwie metody selekcji cech słowa bazujące na analizie składni struktury frazy oraz analizie składni zależności. Badania przeprowadzono przy wykorzystaniu rożnych baz danych. Proponowana metoda ma większą dokładność niż dotychczas stosowane metody: okna i zależności.