项目名称: 词汇化句法分析若干关键技术研究
项目编号: No.61262035
项目类型: 地区科学基金项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 袁里驰
作者单位: 江西财经大学
项目金额: 43万元
中文摘要: 虽然目前词汇化句法分析研究取得了一定的进展,但是其准确率和效率仍然不能满足实际应用的需要,特别是汉语词汇化句法分析,还有许多关键技术有待解决。近年来,句法分析领域出现了将多种技术进行融合的研究趋势,如何整合多项现有的技术,取长补短,将是句法分析领域的一个重要研究方向。因此本课题将对多项技术的融合问题展开研究,在中心词驱动句法分析模型的基础上结合依存语法、配价语法,提出新的句法语义理论和计算模型:(1)提出一种新的词聚类模型和算法,建立基于语义类的句法分析模型,解决数据稀疏问题;(2)在句法分析统计模型中引入语义信息, 既包括语义依存信息,也包括配价结构等语义搭配信息;(3)词性标注在句法分析中起到了至关重要的作用, 本课题提出的句法分析模型将结合词性标注进行句法分析。目标是充分发挥语义信息在句法分析中的作用,解决句法分析和语义计算中存在的关键问题,较大幅度提高句法分析的性能。
中文关键词: 自然语言处理;句法分析统计模型;配价结构;中心词驱动;语义依存关系
英文摘要: The study of lexicalized syntactic parsing has achieved certain progress, but its precision and efficiency yet do not satisfy the need of practical applications, especially there are still many key technologies about Chinese lexicalized syntactic parsing to be solved. In recent years, there has arised a study tendency to merge varied technologies in the field of syntactic parsing, and how to integrate varied existing syntactic parsing technologies and complement each other will be an important trend in the field of syntactic parsing. So this project will study how to merge varied syntactic parsing technologies, and propose new theories and calculation models about syntax and semantics on the base of Head-Driven syntactic parsing models combined with dependency grammars and valence grammars: (1) Propose a new word clustering model and algorithm, build syntactic parsing models based on semantic category and solve the data sparseness problem; (2) Introduce semantic information for statistical syntactic parsing models, the semantic information include semantic dependency and semantic collocation such as valence structure; (3) Part-of-speech tagging plays a vital role in syntactic parsing, the syntactic parsing models proposed in this project will combine part-of-speech tagging with syntactic parsing. The goals of t
英文关键词: natural language processing;statistical syntactic parsing model;valence structure;head-driven;semantic dependency