基于模糊向量空间的文本分类方法  

Document Classification Method Based on Fuzzy Vector Space

作　　者：郑凤萍[1] 刘春雨[2] 

机构地区：[1]大庆石油学院图书馆,黑龙江大庆163318 [2]吉林大学管理学院,吉林长春130022

出　　处：《情报科学》2007年第4期588-591,共4页Information Science

摘　　要：本文针对文本自动分类问题,提出了一种基于模糊向量空间模型和径向基函数网络的分类方法。网络由输入层、隐层和输出层组成。输入层完成分类样本的输入,隐层提取输入样本所隐含的模式特征,将分类结果在输出层表现出来。该方法在特征提取时充分考虑了特征项在文档中的位置信息,构造出模糊特征向量,使自动分类更接近手工分类方法。以中国期刊网全文数据库部分文档数据为例验证了该方法的有效性。Aiming to document classification in data mining, a classification method based on fuzzy vector space model and radial basis function network is presented in this paper. The network includes input layer, hidden layer and output layer. Input layer performs import of samples, hidden layer extracts model characters of samples and output layer presents classification results. The information of its locality in the document is considered while the keywords of model characters are extracted. The classification results of this method are more precise than that of general method because fuzzy eigenvectors are applied. Finally the availability of model and algorithms is proved by classification of some documents in china periodical document database.

关 键 词：数据挖掘 特征提取 神经网络 文本分类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]