文本分类中信息增益算法的改进  

Improvement of information gain algorithm in text classification

作　　者：杨敬妹[1] 王学军[1] 

机构地区：[1]石家庄铁道大学信息科学与技术学院,河北石家庄050043

出　　处：《计算机时代》2013年第9期45-46,49,共3页Computer Era

摘　　要：分析了信息增益方法的不足,并将类内离散度、类间离散度和权重协调因子应用到信息增益算法上,提出了一种改进的信息增益算法。实验表明,该方法在分类效果上与经典算法相比有一定的提高。In this paper the drawback of information gain method is analyzed carefully. An improved information gain algorithm is proposed, based on the application of within-class scatter, between-class scatter and weights coordination factor. The experimental result shows that this method has enhanced the effect of classification to some extent.

关 键 词：特征选择 信息增益 类内离散度 类间离散度 权重协调因子 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]