Chinese word sense disambiguation with variable context window
Yan Ron · Jisuanji gongcheng yu sheji · 2015
Knowledge-based WSD methods exploiting a certain window size as backgrounds of disambiguation are ineffective on account of not considering the effect of disambiguation noise,and a Chinese WSD model with variable context window(CWSDVCW)was proposed.To reduce noise,the context window was adjusted according to the part-of-speech of polysemous word as much as possible to ensure syntactic relation between every word in the context window and polysemous word.Meanwhile,by building sense collocation corpus,the word semantic relevancy computation was further refined.Experimental results show that the result generated using the proposed method is 8.6% higher than the best result generated in SemEval-2007.