基于网格距离的聚类算法的设计、实现和应用  

Design, realization and application of clustering algorithm based on the distance between grids

作　　者：田启明[1] 王丽珍[1] 尹群[1] 

机构地区：[1]云南大学信息学院

出　　处：《计算机应用》2005年第2期294-296,共3页journal of Computer Applications

基　　金：云南省自然科学基金资助项目(2002F0013M)

摘　　要：提出了一种新的基于网格距离的聚类算法。该算法不仅克服了K 代表点算法中必须事先给定K值、难以确定初始代表点、聚类结果的现实意义难以描述等缺点,而且克服了基于网格的聚类算法中要求数据必须在空间密集的缺陷。通过实验验证了新算法的正确性和有效性。This paper presented a new clustering algorithm based on the distance between grids. The new algorithm not only overcomed the shortcoming of K-Medoid algorithm which has much difficulties to suppose K in advance, confirm the initialized points and explain realistically, but also overcomed the shortcoming of clustering algorithm based on grids which requests dense data in spaces. The new algorithm was proved to be correct and efficient by the results of experiments.

关 键 词：数据挖掘 聚类 网格 K-均值聚类 相似度量 内涵知识 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]