语言信息处理技术中的最大熵模型方法

Method of Maximum Entropy Model for Language Processing

1引言 进行汉语处理时经常遇到的问题有:分词、词性标注、语法和语义分析等等.这些自然语言中的问题都可以形式化为分类问题,估计某一类y在上下文x中发生的概率,即p(y,x).在汉语中上下文x的内容可以包括汉字、词、词性等,对于不同的任务上下文的内容也不同.这类问题可以采用统计建模的方法去处理.

As a statistical method. the framework of maximum entropy is efficiently used. In its applications the accuracy is at or near the state-of-the-art. The model is easy to understand, and at the same time it can control subtle features and has reusability. The goal of this paper is to provide a brief description of formalism for the principle of the maximum entropy. And some important algorithms for parameter estimation and feature induction are also introduced.