Chinese Word Sense Disambiguation based on Context Expansion

Word Sense Disambiguation (WSD) is one of the key issues in natural language processing. Currently, supervised WSD methods are effective ways to solve the ambiguity problem. However, due to lacking of large-scale training data, they cannot achieve satisfactory results. In this paper, we suppose synonyms for context words that can provide more knowledge for WSD task, and present two different WSD methods based on context expansion. The first method regards Synonyms as topic contextual feature to train Bayesian model. The second method treats context words made up of synonyms as pseudo training data, and then derives the meaning of ambiguous words using the knowledge from both training and pseudo training data. Experimental results show that the second method can significantly improve traditional WSD accuracy by 2.21%. Furthermore, it also outperforms the best system in SemEval-2007.