基于特征词权值的渔业文本分类研究  

Fishery text categorization method based on feature word weight and support vector machine

作　　者：谷军[1] 何南[1] 

机构地区：[1]大连海洋大学信息工程学院,辽宁大连116023

出　　处：《大众科技》2014年第12期40-42,共3页Popular Science & Technology

基　　金：辽宁省科研计划项目(2010216008);辽宁省高等学校优秀人才支持计划(LR2012024)

摘　　要：渔业文本分类是充分利用渔业信息资源的有效途径。针对中文文献资料的结构特点,提出一种结合特征词权值和支持向量机(Support Vector Machine,SVM)的渔业文本分类方法,利用向量空间模型(Vector Space Model,VSM)构建文本向量空间,并结合特征词权值计算文本特征向量中的各特征项,将构建的文本向量送入SVM进行渔业文本分类。采用中国知网下载的标准文档进行了实验测试,并考察了准确率和召回率两个指标,实验结果表明,文章提出的渔业文本分类方法具有较好的分类效果。Fishery text categorization is an effective way to make full use of fishery information resources. According to the structural characteristics of Chinese literatures, a fishery text categorization method based on feature word weights and support vector machine was put forward. Text vector space was constructed by using vector space model（VSM） and the feature items in every text feature vector were calculated in consideration of feature word weights. Support vector machine（SVM） was employed as the classifier and the standard documents downloaded from CNKI as the test data. The experiments were checked with precision rate and recall rate. The experimental results show that our fishery text categorization method owns satisfactory categorization performance.

关 键 词：渔业信息资源 渔业文本分类 特征词权值 支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]