Collaborative filtering by sequential user-item co-cluster extraction from rectangular relational data

2014 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE)

Oda

Notsu

2014

Self Cite

This paper considers fuzzy co-clustering of distributed cooccurrence data, where vertically partitioned cooccurrence information among objects and items are stored in several sites. In order to utilize such distributed data sets without fear of information leaks, a privacy preserving procedure is introduced to fuzzy clustering for categorical multivariate data (FCCM). Withholding each element of cooccurrence matrices, only object memberships are shared by multiple sites and their (implicit) joint co-cluster structures are revealed through an iterative clustering process. Several experimental results demonstrate the ability of improving the individual coclustering results of each site by combining the distributed data sets.

Section: Discussionmentioning

confidence: 99%

mentioning

confidence: 99%

Fuzzy co-clustering of vertically partitioned cooccurrence data with privacy consideration

2014 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE)

Oda

Notsu

2014

Self Cite

“…Fuzzy variants of co-clustering have also been demonstrated to be useful in such applications as document analysis [8] and collaborative filtering [9,10]. The goal of fuzzy co-clustering is to simultaneously estimate memberships of both objects and items from a cooccurrence information matrix.…”

Section: Introductionmentioning

confidence: 99%

A Collaborative Framework for Privacy Preserving Fuzzy Co-Clustering of Vertically Distributed Cooccurrence Matrices

Advances in Fuzzy Systems

Oda

Tanaka

et al. 2015

Self Cite

In many real world data analysis tasks, it is expected that we can get much more useful knowledge by utilizing multiple databases stored in different organizations, such as cooperation groups, state organs, and allied countries. However, in many such organizations, they often hesitate to publish their databases because of privacy and security issues although they believe the advantages of collaborative analysis. This paper proposes a novel collaborative framework for utilizing vertically partitioned cooccurrence matrices in fuzzy co-cluster structure estimation, in which cooccurrence information among objects and items is separately stored in several sites. In order to utilize such distributed data sets without fear of information leaks, a privacy preserving procedure is introduced to fuzzy clustering for categorical multivariate data (FCCM). Withholding each element of cooccurrence matrices, only object memberships are shared by multiple sites and their (implicit) joint co-cluster structures are revealed through an iterative clustering process. Several experimental results demonstrate that collaborative analysis can contribute to revealing global intrinsic co-cluster structures of separate matrices rather than individual site-wise analysis. The novel framework makes it possible for many private and public organizations to share common data structural knowledge without fear of information leaks.

“…items are forced to be 1 in each cluster, i.e., item memberships only play a role for evaluating the relative responsibility of items in each cluster. This type of co-clustering model was proved to be useful in collaborative filtering tasks [11], [12] and document analysis [13], in which several popular items or keywords can be shared by multiple clusters.…”

Section: Introductionmentioning

confidence: 99%

A cluster validity index for FCM-type co-clustering

Muranishi

2013 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE)

Notsu

et al. 2013

Self Cite

Cluster validation is an important issue in FCMtype clustering applications and many validity indices have been proposed for selecting the optimal fuzzy partition. In most of FCM-type cluster validity indices, the number of clusters is evaluated by considering the partition quality and geometrical features. The quality of memberships was evaluated by the degree of overlapping or the boundary uncertainty, while the geometrical quality was measured by cluster compactness or cluster separation. In this paper, a new approach for FCM-type cluster validation in fuzzy co-clustering is proposed. Because fuzzy co-clustering extracts object-item dual clusters without using prototypes, a new concept of cluster separation is constructed without using the distances between cluster prototypes. The applicability of the new validity index is demonstrated in several numerical experiments including a document clustering application.