结合粗糙集与集成学习的中文文本分类方法研究_张翔.pdfVIP

第 28卷第 1期 计算机应用与软件 Vol128No. 1 2011年 1月 Computer Applications and Sofwt are Jan. 2011 1, 2 3 1 1 张 翔 周明全 董丽丽 闫清波 1 ( 7100 ) 2 ( 710127) 3 ( 10087 ) 介绍中文文本分类的流程及相关技术在分析传统的文本特征 择不足的基础上, 提出了基于粗糙集与集成学习结合 的文本分类方法, 通过粗糙集进行文本的特征 择, 采用 一种集成学习算法AdaBoost. M 1来提高弱分类器的分类性能, 对中文文本 进行分类实验证明, 这种算法分类结果的 F1值比 C4. kNN分类器都高, 具有更加优良的分类性能 中文文本分类 粗糙集 集成学习 AdaBoost. M 1 ON CHINESE TEXT CATEGORIZATION BASED ON ROUGH SET AND ENSEMBLE LEARN ING 1, 2 3 1 1 Zhang X iang Zhou M ingquan Dong L ili Yan Q ingbo 1 (School of Inf orm ation and Control Eng ineering, X i. an University of Arch itecture and Technology, X i. an 7100 , Shaanx ,i China ) 2 (School of Inf orm ation Science and Technology, Northw est University, X i. an 710127, Shaanxi, China) 3 (School of Inf orm ation Science and Technology, B eij ing Norm al University, Bejiing 10087 , Ch ina) Abstract This paper introduces the flow of Chinese text categorisation and the relevant technologies. A text categorisation approach based on the com bination of rough set and ensem ble learning is proposed on the basis of analyzing the d isadvantage of traditional feature selection, the feature selection of the text is executed through the rough set, and an ensem ble learn ing algorithm AdaBoost. M 1 is em ployed to mi prove the categorising perform ance ofw eak class ifier to categorise the Ch inese text. Expermi ent indicates that this method has am ore excellent classifica- tion performancew ith its F1 value of the categorised outcom e higher