基于概念簇的文本向量构建方法

doi:1000-436X(2010)8A-0044-04

通信学报 ›› 2010, Vol. 31 ›› Issue (8A): 44-47.doi: 1000-436X(2010)8A-0044-04

基于概念簇的文本向量构建方法

冯扬,罗森林,潘丽敏,刘莉莉,陈开江

北京理工大学信息与电子学院信息安全与对抗技术实验室，北京 100081

出版日期:2010-08-25 发布日期:2017-07-03
基金资助:
国家 242 计划基金资助项目;北京理工大学基础研究基金资助项目;北京理工大学研究生科技创新基金资助项目

Method of text vector construction based on concept cluster

Yang FENG,Sen-lin LUO,Li-min PAN,Li-li LIU,Kai-jiang CHEN

Information Security and Countermeasures Laboratory,School of Information and Electronics,Beijing Institute of Technology,Beijing 100081,China

Online:2010-08-25 Published:2017-07-03
Supported by:
The National 242 Projects;The Basic Research Foundation of Beijing Institute of Technology;Graduated Student Science ＆ Technology Creative Project of Beijing Institute of Technology

摘要/Abstract

摘要：

为提高文本向量对文本概念的逼近程度，通过将具有相同语法语义特征的词进行聚类，提取概念簇，利用空间变换将文本向量由词空间变换到概念簇空间上来表达文本。实验比较了基于TF-IDF、IG、TF-IDF-IG、LSA以及它们结合概念簇后对文本分类的效果，证明了基于概念簇的文本向量构建方法能提高文本向量对文本概念逼近的准确程度，同时也提高了不同类型文本之间的区分度。

关键词: 中文信息处理, 文本向量, 概念簇, 文本分类

Abstract:

To enhance the performance of the text vector,terms were clustered,which contained similar syntax or semantic feature,to construct concept cluster.The text vector would be transformed from term-space to concept-cluster-space to represent the original text.The experiment compared effects of text classification based on TF-IDF,IG,TF-IDF-IG,LSA,and their combinations with concept cluster.And the results show that,the text vector based on concept cluster improves the accuracy of text concept approaching,and advances the discriminating degree between different types of texts.

Key words: chinese information processing, text vector, concept cluster, text classification

冯扬,罗森林,潘丽敏,刘莉莉,陈开江. 基于概念簇的文本向量构建方法[J]. 通信学报, 2010, 31(8A): 44-47.

Yang FENG,Sen-lin LUO,Li-min PAN,Li-li LIU,Kai-jiang CHEN. Method of text vector construction based on concept cluster[J]. Journal on Communications, 2010, 31(8A): 44-47.

图/表 4

图1

表1

表2

图2

参考文献 13

[1]	SALTON G , WONG A , YANG C S . A vector space model for automatic indexing[J]. Communications of the ACM, 1975,18(11): 613-620.
[2]	CHANGKI L , GARY G L . Information gain and divergence-based feature selection for machine learning-based text categorization[J]. Information Processing ＆ Management, 2006,42(1): 155-165.
[3]	GUO Z L . Using mutual information to identify new features for text documents of various domains[A]. Proceedings of the 17th Asia Conference (PACLIC17)[C]. Sentosa,Singapore, 2003. 372-379.
[4]	DENG Z H , TANG S W , YANG D Q ,et al. A comparative study on feature weight in text categorization[A]. Proceedings of Advanced Web Technologies and Applications 6th Asia-Pacific Web Conference (APWeb 2004)[C]. Hangzhou,China, 2004. 588-597.
[5]	WILBUR W J , YANG Y M . An analysis of statistical term and its use in the indexing and retrieval of molecular biology texts[J]. Computers in Biology and Medicine, 1996,26(3): 209-222.
[6]	DEERWETER S , DUMAIS S T , FURNAS G W ,et al. Indexing by latent semantic analysis[J]. Journal of the American Society for Information Science, 1990,41(6): 391-407.
[7]	鲁松, 李晓黎, 白硕等 . 文档中词语权重计算方法的改进[J]. 中文信息学报, 2000,14(6): 8-13. LU S , LI X L , BAI S ,et al. An improved approach to weighting terms in text[J]. Journal of Chinese Information Processing, 2000,14(6): 8-13.
[8]	沈志斌, 白清源 . 文本分类中特征权重算法的改进[J]. 南京师范大学学报(工程技术版), 2008,8(4): 95-98. SHEN Z B , BAI Q Y . Improvement of feature weighting algorithm in text classification[J]. Journal of Nanjing Normal University (Engineering and Technology Edition), 2008,8(4): 95-98.
[9]	董小国, 甘立国 . 基于句子重要度的特征项权重计算方法[J]. 计算机与数字工程, 2006,34(8): 35-37. DONG X G , GAN L G . Feature weighting based on the importance of sentence[J]. Computer ＆ Digital Engineering, 2006,34(8): 35-37.
[10]	陆玉昌, 鲁明羽, 李凡等 . 向量空间法中单词权重函数的分析和构造[J]. 计算机研究与发展, 2002,39(10): 1205-1210. LU Y C , LU M Y , LI F ,et al. Analysis and construction of word weighing function in VSM[J]. Journal of Computer Research and Development, 2002,39(10): 1205-1210.
[11]	刘里, 何中市 . 基于关键词语的文本特征选择及权重计算方案[J]. 计算机工程与设计, 2006,27(6): 934-936. LIU L , HE Z S . Term selection and weighting approach based on key words in text categorization[J]. Computer Engineering d Design, 2006,27(6): 934-936.
[12]	吴科, 石冰, 卢军等 . 基于文本集密度的特征选择与权重计算方案[J]. 中文信息学报, 2003,18(1): 42-47. WU K , SHI B , LU J ,et al. Feature selection and weighting scheme based on text set density[J]. Journal of Chinese Information Processing, 2003,18(1): 42-47.
[13]	胡和平, 曾庆锐, 路松峰 . 中文词聚类研究[J]. 计算机工程与科学, 2006,28(1): 122-124. HU H P , ZENG Q R , LU S F . Research on chinese word clustering[J]. Computer Engineering ＆ Scienc, 2006,28(1): 122-124.

编号	簇内特征词及其与中心向量的相似度
1	前卫/0.326039 边锋/0.378513 前卫/0.384909 中路/0.328963 中卫/0.250125 前锋/0.398463 门将/0.402236中场/0.351136 射手/0.312999
2	宝来/0.665696 福田/0.323951 华菱/0.287651 老宝来/0.473288 蒙迪欧/0.499464 新宝来/0.599274 广州本田/0.582739 黄海/0.287651 皇冠/0.448904 铃木/0.671791长安/0.551088 嘉年华/0.490048 华普/0.601246
3	外币/0.361532 本息/0.517433 贷款/0.678472 港币/0.55564 微观调控/0.495544 违约金/0.382829 准备金/0.68706 房市/0.535852 恒生/0.510985 借款/0.537883调控/0.42729 资信/0.50078 分红/0.482312 红利/0.449838 净值/18:0.536353 票据/0.424019
4	波长/0.365115 复合材料/0.388535 光通信/0.548256 海量/0.320111 画质/0.266502 接入网/0.5294 接收机/0.436962 光纤通信/0.283583 零件/0.43516 扩容/0.375232 存储器/0.518526
5	南联盟/0.389521 地空导弹/0.336192 截击机/0.320587猛禽/0.632521 萨姆/0.444748 制空权/0.459399 警报/0.36304 螺旋/0.271911 炸药/0.180785 机群/0.336785掩体/0.403364 防区/0.518854 中巴/0.389537

方法		IT	军事	汽车	体育	平均
TF-IDF	P	0.844	0.875	0.874	0.951	0.886
	R	0.802	0.927	0.874	0.937	0.885
TF-IDF +概念簇	P	0.945	0.953	0.962	0.973	0.958
	R	0.904	0.967	0.964	0.992	0.957
IG	P	0.846	0.878	0.877	0.959	0.890
	R	0.803	0.928	0.881	0.939	0.888
IG +概念簇	P	0.938	0.954	0.946	0.972	0.953
	R	0.887	0.963	0.967	0.988	0.951
TF-IDF-IG	P	0.898	0.938	0.891	0.964	0.922
	R	0.826	0.948	0.940	0.967	0.920
TFIDFIG+概念簇	P	0.915	0.959	0.926	0.969	0.942
	R	0.887	0.947	0.947	0.984	0.941
LSA	P	0.867	0.897	0.878	0.972	0.904
	R	0.812	0.945	0.911	0.937	0.901
LSA +概念簇	P	0.946	0.955	0.961	0.973	0.959
	R	0.905	0.967	0.964	0.995	0.958

基于概念簇的文本向量构建方法

Method of text vector construction based on concept cluster

在线阅读

PDF下载

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 4

参考文献 13

相关文章 3

Metrics

推荐阅读 0

[1]	刘浩然,丁攀,郭长江,常金凤,崔静闯. 基于贝叶斯算法的中文垃圾邮件过滤系统研究[J]. 通信学报, 2018, 39(12): 151-159.
[2]	王博,贾焰,杨树强,周斌. 适用于不确定文本分类的特征选择算法[J]. 通信学报, 2009, 30(8): 32-38.
[3]	杨震,范科峰,雷建军,赖英旭. 基于分类器集成的拉圾邮件过滤方法研究[J]. 通信学报, 2008, 29(11A): 7-11.