| Word sense disambiguation (WSD) has been an important research subject in semantic study and Natural Language Processing which covers serious of natural language processing system including: message retrieval, machine translation, key word extraction, speech recognition, text categorization and automatic abstract. Linguists and computer scientists have contributed to WSD and gained distinguished achievement. However, the existed researches on WSD mostly focus on grammar, dictionary, and ordinary meanings of simple words, but pay limited attention on the factors of context, and pragmatics. Moreover, there is no attempt on the WSD of modal verbs in Chinese, whose meanings are fuzzy and sensitive to contexts.Since language is the object of word sense disambiguation, application of research findings on linguist could break a great through in WSD. Besides, since the publication of Ma's Grammar (Ma, 1898), linguists have made a comprehensive study on Chinese modal verbs, which is stable foundation for the model construction of WSD model for Chinese modal verbs.On the other hand, the evolution of WSD will improve the development of linguistic study. With the successful of automation of sense disambiguation, each modal verb can also be tagged automatically in corpus, which can save amount of time and energy for scientists who study modal verbs based on corpus. Therefore, this study will contribute to the linguistic study practically and theoretically.Based on connectionism, this thesis presents a disambiguation model of Chinese modal verb'能'(néng, can) with a method of Neural Network which features in self-organization, self-adoption and high fault tolerance. A relatively high accuracy as 96% is obtained and the deep relationship between modal verb'能'(néng, can) and its contextual features are figured out.
|
PDF Full Text Request