一种基于群体增量学习算法的文本特征选择方法

A Text Feature Selection Method Using the Population Based Incremental Learning Algorithm

尽管目前存在许多文本特征选择方法,但是它们都有着一定的局限性。提出一种新的基于群体增量学习(Population Based Incremental Learning)算法的文本特征选择方法,其特点是无需特征集的先验知识和容易实现,并且由于使用了简单分类器性能作为评价准则,计算复杂度很低。对Reuters-21578文本集的分类实验结果表明,该方法平均分类性能要优于卡方统计量、信息增益和简单遗传算法三种常用的特征选择方法。

At present there are many methods to deal with text feature selection, but each of them has certain disadvantages. A novel text feature selection method using the population based incremental learning algorithm is introduced in this paper. Advantages of the proposed method are that it needs no priori knowledge of features, is easily implemented and its computational complexity is very low due to using a simple classifier. Experimental results obtained from the Reuters - 21578 dataset show that the method is better than chi - square, information gain and genetic algorithm in the performance of text categorization.