期刊名称:International Journal of Computer Science and Network Security
印刷版ISSN:1738-7906
出版年度:2017
卷号:17
期号:1
页码:67-72
出版社:International Journal of Computer Science and Network Security
摘要:Fuzzy co-clustering is a basic tool for extracting pair-wise clusters of familiar objects and items from cooccurrence information. A promising improvement of the conventional fuzzy co-clustering algorithms is achieved by introducing exclusive nature to item partition with the goal of the improvement of interpretability of co-clusters. However, in practice, some items are quite popular and to be shared by multiple clusters, and only a selected part of items should be exclusively assigned to unique clusters. In this paper, a partially exclusive item partition model is introduced into multinomial mixture models-induced fuzzy co-clustering and a two phase implementation is proposed for determining the optimal set of items to be exclusively assigned. Its characteristic features are demonstrated through a numerical experiment with a real-world benchmark data set.