R-means：以关联规则为簇中心的文本聚类

本文将k-means与关联规则（或频繁项目集）相结合,提出了一种新的文本聚类算法R-means.R-means算法以关联规则作为簇中心,通过类似于k-meams的迭代优化得到最终的簇.因此R-means不仅继承了k-means的简单性,而且用关联规则产生的簇描述易于为人们所理解.在几个实际数据集上的实验表明该算法可以得到高精度和高性能.

R-means: Exploiting Association Rules as Means for Text Clustering

This paper proposes a new text clustering algorithm called R-means which integrates k-means with associa- tion rule (or frequent itemset). R-means exploits association rules as means of clusters and refines clusters by an itera- tive procedure which is similar to that of k-means. R-means not only inherits the simplicity of k-means, but also gener- ates more comprehensive cluster labels which are described by association rules. The experiments with several real data sets have demonstrated that R-means can achieve quite well precision and high performance.