面向语言信息智能接口的线性文法研究

Chinese information processing indubitably need to utilize Chinese grammar roperties. However, the system is basically copied from English. In addition, it is not suit computers but people who are studying Chinese and it is not suite the word neighborship processing but structure analysis. To improve performance of Chinese processing applications, we must change this situation, thoroughly think of the linguistics foundation of Chinese processing, and work over grammar that suits to process word neighborship in Chinese by computers. That is why we are studying Chinese linear grammar..We did research on the part-of-speech systems that are currently influential in the Mainland and Taiwan, and investigated the shortages in intelligent interface system of language information due to statistic dada sparseness. We tagged meta information of texts in a large scale Chinese novel corpus containing about one hundred million characters, developed the first retriever in the world for word attributes from Chinese raw corpus, improved and expanded the capacity of the general purpose word segmentation system GPWS, and tried to discover which word attributes required in the applications related to language processing. Based on all these works, as an original innovation, we brought forward the accidence design of linear grammar system based on word attributes. Research on this grammar system will lead to reestablishment of the linguistics foundation of the Chinese information processing and bring the distinct improvement in performance of the applications..

本项目研究面向汉语语言信息智能接口的线性方法，特别是适合于经性文法的汉语词类系统，以及使用词和词类为语言单位和统计语言模型。本项目的研究成果将缓解线性文法中数据稀疏的，提高汉语语言信息智能输入输出系统的准确性，推动我国计算机和网络的普及，推进我国经济和信息化的进程