Subcategorization Driven Chinese Syntactic Analysis

Parsing is one of the crucial problems in natural language processing. The goal of parsing is to derive the syntactic structure of a sentence according to a certain grammar. The improvement of parsing will give an enormous impetus to natural language processing applications such as machine translation, information retrieval, information extraction and speech processing. Chinese syntactic parsing technique is the basis for Chinese information processing. The grammatical characteristics of Chinese causes us unable to copy the traditional methods applied to the analysis of English parsing. We should proceed from the characteristics of Chinese to study the techniques suitable for the analysis of the Chinese parsing. The study of Chinese parsing will help provide various solutions for problems of natural language processing. It is one of the most important means to solve the natural language understanding.Firstly, this dissertation introduces the related theory and technology of syntactic analysis and existed parsing algorithms and models. And it makes a brief introduction of the sub-categorization principle as well as the relationship between the analysis of Chinese parsing and the sub-categorization of verbs. This dissertation proposes the method for syntactic analysis based on the sub-categorization rules and method for Chinese parsing based on the statistical characteristics of sub-categorization. We bring the conception of the most similarity into the regular methods of the analysis of parsing and formulate the matching methods of regular models by means of using the information of sub-categorization of verbs. This dissertation also points out the conceptions of average probability based on the statistical parsing methods and analyzes the Chinese sentences by dint of matching and getting the average probability of the rules of parsing and sub-categorization rules. The results we've obtained from the two above methods makes considerable improvements when compared with the basic ones.