| Summary: | 碩士 === 淡江大學 === 資訊工程學系 === 90 === In the recent ten years, with the development of network technology, it makes growth of internet population. We can not only get news and interested information but also shopping and transacting some formalities via the internet. However the website lacks for active and mutual interactive mechanism, so it lets users spend much time on internet but still not find anything they want. Therefore the way of understanding and representation what users’ want is the key factor in deciding the accuracy of responsion. The natural language is the best instrument that can represent the thinking of human completely.
There are still some problems for computers to understand natural language. In Chinese language there are two difficulties. First, Chinese language has no space to segment the lexical entry, so the segmentation method of Chinese is more difficult than English. Second, the lack of complete grammar in Chinese language makes parser more difficult and complicated. These difficulties should be overcome when building a Chinese interactive feedback system. In this thesis, we proposed a method to parse Chinese language, that uses particular lexical database according to the particular application to improve the response time and accuracy.
This thesis begins with the lexical entries that have been segmented by segmentation system. When the system gets segmented lexical entries, it goes through syntactic analysis and semantic analysis to get the structure and meaning of the sentence. Because of the change of grammar and semantic restriction, the system provides fault toleration and learning mechanism for suiting individual oral habit of users. Finally, we encode the lexical entries and take it to search the best response by weighted calculation.
|
|---|
