多核SVM文本分类研究  

Research on the Text Classification Based on Multi-kernel Support Vector Machine

作　　者：陈海红[1] 

机构地区：[1]湖南永州职业技术学院计算机系,湖南永州425000

出　　处：《软件》2015年第5期7-10,共4页Software

基　　金：国家社科基金项目"自动文本分类技术研究"(08CTQ003)资助

摘　　要：支持向量机是在统计学习理论基础上发展起来的一种机器学习方法,广泛应用于文本分类领域。利用信息增益法进行文本特征选取,运用TF-IDF进行特征权重设置。对支持向量机不同核函数,通过网格与交叉验证组合法优化参数选择,比较支持向量机在不同核函数下文本分类性能,得出一些有价值的结论。Support vector machine is one of the machine learning methods based on statistical learning theory. It is widely used in the field of text classification. The text features were chosen with the information gain method, and the feature weight was set based on TF-IDF method. K-cross validation and grid search were combined to optimize the support vector machine (SVM) kernel function’s parameters. Finally, some helpful results were obtained by comparing text classification performance of the SVM under different kernel functions.

关 键 词：文本分类 支持向量机 核函数 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TP391.1[自动化与计算机技术—控制科学与工程]