Chinese Word Sense Disambiguation Based On Moses

The use of ambigous vocabulary brings great changes and much convenience for Chinese language.Meanwhile,it also brings great difficulty for natural language processing(NLP).The task of word sense disambiguation(WSD)is to find true sense of ambiguous word based on specific language environment.WSD is a common and difficult problem in NLP.With the development of NLP,WSD has become a most pressing and fundamental issue.With in-depth study of WSD,this paper proposes a WSD method based on statistic studying theory.It considers Chinese sentence as a processing unit and attempts to find a corresponding sense for each vocabulary in sentence.The method can make good use of knowledge in language environment.It is flexible and it can adapt to the development of language.There are three aspects of research work in this paper.Firstly,this paper states emergence and research significance of WSD,and it shows the development of WSD at home and abroad.Some application scenarios of classical WSD method are explained and some problems which can occur in WSD are analyzed in this paper.Secondly,it investigates corpus and corresponding dictionary.The aspect of corpus contains its origin,organization form and its preprocessing.The aspect of dictionary contains the correlation knowledge of word sense in Tong Yi Ci Ci Lin.All of sense information in experiment is denoted by three levels' coding of word sense in “Tong Yi Ci Ci Lin”.Thirdly,the paper shows the origin of used knowledge for WSD.The knowledge of phrase-semantics and the semantic correlation knowledge between contiguous words are used to describe language invironment of ambiguous word.Statistic studying theory is employed to learn the knowledge and build WSD model.Moses decoding algorithm is adopted to find an optimal sequence ofsemantic categories for Chinese sentence in its semantic nework,so true sense of ambiguous word can be decided.Finally,contrast test is used to evaluate the performance of WSD model.