Research On Chinese Syntactic Parsing Based On SEARN Algorithm

Syntactic parsing plays a critical role in natrual language processing and has been drawing more and more attention due to its importance in deep natrual language processing applications, such as information extraction, question answering, and machine translation. So many researchers focus on it. Due to the difficulty in Chinese syntactic parsing, this paper presents a novel approach which applys the Shift-Reduce Chinese paring in SEARN algorithm.In details, the main content of this paper includes:1) The research on syntactic parsing methods based on statistical learning models. In recent years, much attention has been focused on the problem of syntactic parsing, and a variety of approaches based on statistical learning models have been proposed. In this paper, we systemically summarize and classify various approaches to syntactic parsing from the view of the adopted statistical learning models and algorithms, focusing on the analysis and comparison of the different types of models and algorithms. The current research on the Chinese syntactic parsing is also presented in this paper. Finally we give the future directions and trends in syntactic parsing research, especially for Chinese syntactic parsing.2) The research on Shift-Reduce parisng. Although Shift-Reduce method achieves better performance in Chinese syntactic parsing, it can be improved in many aspects, such as the decoder and feature templates. Based on the present study, this paper applys the BeamSearch and new feature template in Shift-Reduce method, and they improve the performance.3) The research on syntactic parsing based on SEARN. Traditional syntactic parsing methods are based on the PCFG, but the Shift-Reduce method is based on the classifier. The syntactic parsing is essentially a structured prediction problem and not equal to the classying problem. In this paper, by applying SEARN algorithm, will make up for the lack of the Shift-Reduce parsing and improve the performance.The major contribution of the paper is applying SEARN in Shift-Reduce parsing. The experimental results show that the methods we have proposed are effective on Chinese syntactic parsing, outperforming current methods.