Word Sense Disambiguation Based on Semantic Knowledge

Word sense disambiguation (WSD) is an important and hot research topic in machine translation and information retrieval. A new WSD method is proposed in this paper. Chinese sentence containing ambiguous word is segmented into words. Its left and right words’ semantic categories are used as disambiguation features after Tongyici Cilin is consulted. Based on discriminative features, bayesian model is applied to select correct semantic categories for ambiguous words. Training data set is used to optimize bayesian model and test data set is utilized to test the performance of WSD classifier. Experiments show that accuracy of WSD classifier is improved.