关键词: 文本分类, 中心分类法, 经验风险最小化

Abstract:

Empirical risk minimization inductive principle and gradient descent method were used to fix classcentroidvectors in traditional centroidbased text classification algorithms so as to improve the poor expression ability of classcentroidvectors in traditional centroidbased text classification algorithm caused by ignoring the weighting factors of training texts. Then, an improved centroidbased text classification algorithm was obtained, the performance of which is as well as those of support vector machines. Experimental results show that the method adopted in this article is an effective mean to improve the performance of traditional centroidbased text classification algorithms.

Key words: text classification, centroidbased text classification algorithms, empirical risk minimization