基于相异度矩阵的混合属性数据流聚类算法

Novel algorithm for clustering heterogeneous data stream based on dissimilarity matrix

数据流的聚类是数据流挖掘的一个重要问题。提出一种针对混合属性的数据流聚类算法,它采用相异度来代替普通的聚类距离,并将等价相异度矩阵引入聚类过程。基于真实数据集的实验表明该算法比基地同类算法具有更好的聚类性能。

Data stream clustering is an important issue in data stream mining.In this paper,a novel algorithm is presented for clustering data stream with heterogeneous attributes.It adopts dissimilarity instead of the common clustering distance,and an equivalent dissimilarity matrix is used in the clustering process.Then the empirical evidence of this algorithm＇s superiority over CluStream and HCluStream algorithms on the real data sets is given.