基于最大熵的文本分类算法的改进

基于最大熵模型的文本分类算法对不同测试文档的训练结果相差较大.利用Boosting机制改进基于最大熵模型的分类算法,以提高该分类算法的稳定性.实验结果表明,该改进方法可以有效改善基于最大熵模型分类算法的稳定性,且分类精度也有一定的提高.

Improvement of text categorization algorithm based on maximum entropy

The text categorization algorithm based on maximum entropy model is a kind of effective method,and it has better performance than Bayes,KNN,SVM and etc.,which are the typical text categorization algorithms.But it has different training results to different testing documents,that is,the stability of it is worse.For this reason,the algorithm is improved using boosting mechanism in order to advance its stability.Experimental results show that the improving method is valid in improving the stability and the classification accuracy of the text categorization algorithm based on maximum entropy model.