基于类别的特征选择算法的文本分类系统

目前的索引词选择算法大多是基于词频的，没有利用训练样本中的类别信息，为此提出了一种新的基于类别的特征选择算法。该算法根据某个词是否存在于文档中导致该类文档相似度的区别，来确定该词区分不同文档的分辨力，以此分辨力作为选取关键词的重要度。以该算法为基础，设计了一个英文文本自动分类系统，并对该系统进行了测试和结果分析。

An automatic text classifier of class-based feature selection algorithm

Current feature selection algorithms are all based on term frequency,and ignore the class information in the training sample set.A new feature selection algorithm based on class information was put forward.The principle of the algorithm is as follows: according to the similarity difference caused by whether or not a word existed in a document,the discriminative power with that this word distinguished different documents could be determined.And then,the discriminative power was taken as the importance for