聚类分析中的相似度研究

The Similarity Measure in Clustering

针对聚类分析中相似度进行研究。首先,通过分析相似度的构造,将相似度划分为比较、综合及转换的过程,从而提出一般相似度的概念。其次,以一般相似度为基础上,分析了常见相似度的权重分配,并研究多类型混合数据的相似度计算策略。

This paper focus on the similarity measure in clustering. Firstly, through analyzing the computation of similarity measures, the general similarity measure is proposed, which considers the process of calculating similarity to be three steps：comparison, synthesis, and transformation. And then, on the basis of the general similarity measure the weights distribution of some common similarity measures are investigated, and the meth- ods for computing mixed-type data are also discussed.