在文本挖掘中，文档通常以特征向量的形式表示。为了提高文本挖掘算法的运行速度，降低占用的内存空间，过滤掉不相关或相关程度低的特征，提出一种改进的特征选择算法，该算法对特征进行综合考虑，从而更加准确地选取有效的特征。实验验证了改进算法的可行性和有效性。-In text mining，documents are usually

meanwith the form of the eigenvector．In

order to enhance the operating speed and reduce the memory space occupied　and　filter out irrelevant or lower degree

of features，an improved feature selection algorithm　is　presented．Wecomprehensively

considered features，SOfeatures can be more effective andaccurate selected．And

the algorithm is verified exactly with

the instance in this paper．