基于Multi-Agent的社会网络文体分类方法报告方案.doc

基于Multi-Agent理论的社会网络文体分类方法( 文 菡1,王 杨1,2 (1.安徽师范大学 数学计算机学院 安徽芜湖 241000) (2.中国科学技术大学计算机学院237002） 关键词：社会网络，文体识别、Multi-Agent，文体分类 摘 要：由于当前社会网络中的文体分类存在提取困难、缺乏分类方法等问题，很多传统的分类方法在对于社会网络文体的分类上并没有很好的效果。根据中文网络文体的特点，以一种新的概念来代替传统的分类做法，与传统的方法不同，论文从特征提取的角度提出了一种基于Multi-agent的属性融合和词库关联的网络文体分类方法。首先利用中国科学院计算技术研究所的ICTCLAS系统平台所提供的句法语义处理模块及语料资源，直接提取网络文体的特征关键词和词义等基本属性；然后利用智能Agent理论中的语义特征描述方法，建立基于支持向量机的网络文体识别方法。通过与传统方法的比较实验, 证明本文提出的（基于Multi-agent的）网络文体识别方法能有效地提高文本分类正确率, 同时方法简便，效果良好。 中图分类号：TP391 文献标识码：A 文章编号： Social network style classification method Based on Multi-Agent Wen Han2,Wang Yang1, 2 (1. School of Mathematics Computer Science, Anhui Normal University, Wuhu, Anhui 241000, China; (2. School of Computer Science and technology, University of science and technology of China, Wuhu, Anhui 241000, China) E-mail:wycap@ustc.edu.cn Abstract: Because the stylistic classification and extract the difficulties in the current social network, the problem such as lack of classification method, a lot of the traditional classification method on the stylistic classification for the social network is not very good effect. According to the characteristics of the network in Chinese style, with a new concept instead of the traditional classification method, different from the traditional methods, the paper put forward from the Angle of feature extraction based on Multi - agent of stylistic classification and thesaurus associated network attribute fusion method. First use of computing technology, Chinese academy of sciences institute of ICTCLAS system platform provided by the syntactic and semantic processing module and the corpus resources, direct extraction of network basic properties such as stylistic meaning and characteristics of keywords; Then use description method of semantic features in the theory of intelligent Agent based network style recognition method based on support vector machine (SVM). By comparative experiments with traditional methods, prove that the proposed network styl