Based On Probabilistic Context-free Grammar Parsing And Realization

This dissertation discusses the theory and realization of the Probabilistic Context Free Grammar-based syntactic parser. At first, this paper looks back the history of the natural language processing and its application fields and illustrates that syntactic parser has very importance position of the syntax parser in the natural language processing. At the same time, the main work in this dissertation is given. In the second section, it mainly discusses the syntactic parsing in common use. The chart parser has a lot of advantages, so this method is used in our syntactic parser and is detailed presented. Our syntactic parser is based on Probabilistic Context Free Grammar, for this mode is the better one in the syntactic pattern. In the third section we discuss the method how to remove the ambiguity with PCFG and use an example to explain clearly. The most important thing in the PCFG is that how to gain the probability. In order to solve this problem, this dissertation point out tree algorithms in the fourth section and how to solve the three basilica problems in PCFG. For example, when a Probabilistic Context Free Grammar and the sentence are given, we use the Inside Algorithm to pick-up the probability of the rule; when we want to acquire the best syntactic parsing tree, we gain the probability by using Viterbi Algorithm; at last, we use Inside-Outside Algorithm to choose the probability of the rule for the grammar in order to gain the maximal probability of the trained sentences. We may meet a lot of problems, when we deal with the real sentences with the syntactic parsing. There points out many resolve methods aiming at some particular difficulties in the fifth section. The main aspects are as follows. 1) Chinese have little form changed and the flexibleness of the language's order, so I utilize the phrase-based mode. To some extent, it can get over disadvantages in Chinese. 2) In my syntactic parser, I design a front processing system for the particular character of the Chinese. We can utilize special words in Chinese to forecast the sentence's structure before the synthesis parsing, this is to say, the front processing system actually is the part-syntactic parsing. The main function is to induct the synthesis parsing and reduces a great deal of calculations quantity. 3) I use the data smoothing technology to solve the data sparse problem in the statistic-based syntactic parsing and achieve some improvements. 4) There always have special relations between words in Chinese, so I use this...