基于最大熵模型的中文文本层次分类方法  

Hierarchical Text Categorization Methods Based on Maximum Entropy Model

作　　者：肖雪[1] 

机构地区：[1]重庆电子工程职业学院,重庆401331

出　　处：《计算机与网络》2015年第9期36-38,共3页Computer & Network

摘　　要：针对文本信息海量增加的现状,快速、准确、全面地获取有用信息的大规模信息处理应用技术越来越受到关注。本文将中文文本分类的类别体系构建为层次结构,并把最大熵模型引入中文文本的层次分类,该模型用于得到未知事件分布的最大熵。实验证明,最大熵模型方法的层次分类性能在很多时候优于平面分类,是一种有效的中文文本分类方法。In view of the present situation of mass text information, the technology of large-scale information processing and application, with which people can obtain useful information quickly, accurately, and comprehensively, draws more and more attention. This paper organizes categories into hierarchical structure according to the certain relations. And the maximum entropy model is introduced to hierarchical text classification and used to obtain the maximum entropy of unknown event distribution. The experimont~ results show that the hierarchical classification performance of maximum entropy model outperforms that of plane methods, and it is an effective technique for text classification.

关 键 词：文本分类 层次分类 特征选择 最大熵模型 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]