基于组合凸线性感知器的文本分类模型-JournalofSystems.PDF

第43卷 第11期 北 京 工 业 大 学 学 报 Vol.43 No.11 2017年 11月 JOURNAL OF BEIJING UNIVERSITY OFTECHNOLOGY Nov. 2017 基于组合凸线性感知器的文本分类模型 李玉鑑,王曼丽,刘兆英 (北京工业大学信息学部,北京摇 100124) 摘摇 要:针对文本分类问题,从分片线性学习的角度出发,提出了一种文本分类的组合凸线性感知器模型. 首先, 对文本样本集进行预处理,包括特征选择、特征项赋权等;然后,分别利用生长支持组合凸线性感知器算法(growing support multiconlitron algorithm,GSMA)和支持组合凸线性感知器算法(support multiconlitron algorithm,SMA)构造组 合凸线性感知器,对样本集进行分类. 该模型基于支持向量机的最大间隔思想,通过集成线性分类器,实现了对2 类数据的划分,具有计算简单、适应能力强的优点. 在标准文本数据集上的实验结果表明:该模型所构造的分类器 具有良好的文本分类性能,与其他典型文本分类方法的对比也说明了该方法的有效性. 关键词:分片线性分类器;组合凸线性感知器;文本分类 中图分类号:TP181 文献标志码:A 文章编号:0254-0037(2017)11-1681-09 doi:10.11936/ bjutxb2016070005 Text Classification Model Based on Multiconlitron LI Yujian,WANG Manli,LIU Zhaoying (Faculty of Information Technology,Beijing University of Technology,Beijing 100124,China) Abstract:To deal with the problem of text classification, a text categorization method was proposed based on multiconlitron from the perspective of piecewise learning. First,text sample preprocessing including featureselectionandfeatureweightingwasperformed. Then,themulticonlitronwasconstructed by using growing support multiconlitron algorithm (GSMA) and support multiconlitron algorithm (SMA) respectively for text classification. Inspired by the idea of maximum interval of support vector machine, the classification of two kinds of data by integrating the linear classifier was achieved by this model, which had the advantages of small computation cost and strong adaptive ability. Experimentson standard text data sets show that the proposed method has a good performance on text classification and the comparison resultswith some other typica