基于λ-Warshall算法的聚类方法  

Clustering method based on λ-Warshall algorithm

作　　者：李刚[1] 刘宏兵[1] 冯岩[1] 

机构地区：[1]信阳师范学院计算机科学系,河南信阳464000

出　　处：《计算机工程与设计》2008年第8期1903-1904,1909,共3页Computer Engineering and Design

基　　金：河南省科技计划基金项目(0624220057);信阳师范学院青年骨干教师计划基金项目(20060503)

摘　　要：根据数据之间的相似性,提出了一种基于改进Warshall算法的数据聚类方法。该方法在传统Warshall算法的基础上,引入聚类因子,构造模糊相似关系的传递闭包。由于相似性的自反性与对称性,该传递闭包就是模糊相似关系的等价闭包,把等价数据分到一类形成聚类。实验结果表明,该方法可得到与传统的K-均值聚类算法相同的聚类结果。Clustering method based the improved Warshall algorithm is proposed by using the similarity of data. The method on the basis of the traditional Warshall algorithm forms the transitive closure of fuzzy similar relation by using the clustering parameter. The transitive closure of fuzzy similar relation is the equivalence closure owing to the reflexive and symmetric properties of fuzzy similar relation. The data with the equivalence relation are divided into the same clusters. The experimental results show that the clustering algorithm achieves the same cluster results comparing with the traditional k-means clustering algorithm.

关 键 词：模糊关系 模糊关系矩阵 WARSHALL算法 λ-Warshall算法 聚类 传递闭包 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]