作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

›› 2005, Vol. 31 ›› Issue (13): 13-15.

• 论文 • 上一篇    下一篇

基于《知网》信息结构模式的词汇化树邻接文法消歧

许云,樊孝忠,张锋   

  1. 北京理工大学计算机科学与工程系,北京理工大学计算机科学与工程系,北京理工大学计算机科学与工程系 北京 100081
  • 发布日期:2005-07-15

Lexicalized Tree Adjoining Grammar Disambiguation Based on Message Structure Pattern of HowNet

XU Yun, FAN Xiaozhong, ZHANG Feng   

  1. (Department of Computer Science and Engineering, Beijing Institute of Technology, Beijing 100081)
  • Published:2005-07-15

摘要: 针对词汇化树邻接文法的歧义现象,利用《知网—信息结构库》作为知识库,并引入概率方法提出了一种有效的句法树消歧方法。文中阐述了如何利用知网信息结构模式对句法树进行筛选,并提出了基于概率的信息结构抽取的计算方法。

关键词: 词汇化树邻接文法, 消歧, 知网, 信息结构模式

Abstract: Considering the disambiguation of syntactic tree generated by lexicalized tree adjoining grammar, a statistical approach efficiently solves the disambiguation of syntactic tree using message structure pattern of HowNet as knowledge database. The method to select a optimal syntactic tree based on message structure pattern, and a statistical approach of extracting message structure are presented.

中图分类号: