Abstract: The hierarchical clustering is an important method of clustering analysis. This kind of method can decompose the datum into different levels, and support people with an understanding procedure from superficiality and generality to detail and specific. A new hierarchical clustering method based on GiST-tree is proposed, which can store the structure of the tree generated during the clustering procedure into the hard disk. So it can support very detailed analyzing procedure. The users can discover the relationship among clusters conveniently with this method.

Key words: GiST, hierarchical clustering, data mining, object-oriented technology

摘要： 层次聚类方法是聚类分析的一个重要方法。该文利用通用搜索树实现了一种新的层次聚类算法，可以把整个聚类过程中形成的树型结构都保存在硬盘上，支持从宏观到细微的分析过程，便于用户发现各个聚类之间的相互联系。

关键词: 通用搜索树, 层次聚类, 数据挖掘, 面向对象技术

CLC Number: 

• TP311