摘要
语料库方法在词性标注上获得了较大的成功,但句法分析中仍存在许多问题.针对句法分析方法的不足,文中给出了一个基于语料库的动态规划分析模型.其算法按自底向上的方式逐层构造各种句法树.它可以像枚举分析那样,从所有可能的句法树中选择最合理的句法结构,还可以将复杂度控制在多项式范围内.作为比较,还详细讨论了基于语料库的枚举分析方法。
Corpus based method has been widely applied in natural language processing and satisfactory result has been obtained in Part of Speech tagging; But there still exist many problems in syntactic analysis based on the method. This paper gives a dynamic programming algorithm for corpus based parsing. This algorithm can construct all kinds of syntactic trees from bottom to top step by step, and choose the best one from them in polynomial time, like enumeration method, by which, optimal one is obtained in exponential time! This paper also discusses corpus based enumeration algorithm and its complexity in detail.
出处
《计算机学报》
EI
CSCD
北大核心
1999年第10期1019-1024,共6页
Chinese Journal of Computers
基金
国家自然科学基金
关键词
动态规划
自然语言处理
语料库
句法结构
Syntactic tree, enumeration, dynamic programming, complexity.