摘要
主题概念抽取是文本自动处理中的一项重要工作。以往主题概念抽取中的加权算法较少考虑到主题概念间的相关信息,在一定程度上影响了主题概念抽取的质量。该文提出了一种基于关联矩阵的主题概念选择算法。该算法在概念语义关联矩阵的基础上,通过对矩阵中概念相关向量与文本向量距离的计算,得出候选主题概念相对于待标引文档的重要度,最后依据该重要度完成文本主题概念的选择。实验显示,该算法产生的自动标引结果比单纯按权重排序的方法更能表现文本的主题。
<Abstrcat>Subject concept distillation is an important task in text information automatic processing. In the past, we seldom considered the information related subject concepts in term weighting when extracting subject concepts from texts,which restricted subject concept distillation to some extent . In this paper, an algorithm about choosing subject concepts based on incidence matrix is presented. According to computing distances between the dependence vectors and the document vectors in the semantic incidence matrix, the importance between subject concept and document is put forward. At last, the subject concepts are chosen by the importance. Our experiments show that the indexing results using this algorithm are much more reasonable and acceptable than those using a method purely based on weighting and ordering, and the results are much closer with those of manual work.
出处
《计算机仿真》
CSCD
2005年第5期90-91,119,共3页
Computer Simulation
基金
863计划资助项目(2202AA119050)。
关键词
主题概念
关联矩阵
相关度
Subject concept
Incidence matrix
Relevancy
