An Approach to Microscopic Clustering of Terms and Documents

An Approach to Microscopic Clustering of Terms and Documents
复制标题

术语和文档微观聚类的方法

DOI:
10.1007/3-540-45683-x_44
复制
发表时间:
2002
期刊:
--
影响因子:
--
通讯作者:
Akiko Aizawa
Akiko Aizawa
中科院分区:
--
文献类型:
--
作者:
Akiko Aizawa

文献摘要

被引文献

相似文献

在本文中,我们提出了一种基于文本的信息检索系统中的聚类方法。所提出的方法产生重叠的集群,其中每一个是由相关的条款和规范化的重要性权重的文件的子集。在本文中,我们首先简要介绍了我们的聚类方案的概率制定,然后显示集群生成的过程。我们还报告了一些实验结果,其中在自动文本分类的框架下研究了生成的聚类。
In this paper, we present an approach to clustering in text-based information retrieval systems. The proposed method generates overlapping clusters, each of which is composed of subsets of associated terms and documents with normalized significance weights. In the paper, we first briefly introduce the probabilistic formulation of our clustering scheme and then show the procedure for cluster generation. We also report some experimental results, where the generated clusters are investigated in the framework of automatic text categorization.