A partitioning based algorithm to fuzzy co-cluster documents and words

A heuristic-based fuzzy co-clustering algorithm for categorization of high-dimensional data

“…maximizing the degree of aggregation leads to the formation of highly coherent co-clusters as discussed in Section 2.1). However, due to the bias, such directions may not be able to lead to the desired outcome [16,11]. At this point we note, by observing the membership constraints in Eqs.…”

“…There are a number of existing fuzzy co-clustering algorithms in the literatures, including FCCM [11], Fuzzy CoDoK [7], Fuzzy Simultaneous KeyWord Identification and Clustering of text documents (FSKWIC) [5], and Fuzzy Co-clustering with the Ruspini's condition (FCR) [16]. As mentioned in Section 1, FCCM and Fuzzy CoDoK are two prominent fuzzy co-clustering algorithms based on the partitioning-ranking approach.…”

“…We then go through the detailed formulation of HFCR. Lastly, we explain why there is a need for HFCR, despite our earlier effort in [16] to devise FCR, a fuzzy co-clustering algorithm based on the dual-partitioning approach that can be regarded as the predecessor of HFCR.…”

“…As mentioned in the beginning of this section, we have earlier introduced a fuzzy co-clustering algorithm, called FCR, based on the dual-partitioning approach [16]. Here we explain why, despite FCR, we need to formulate a new dual-partitioning-based fuzzy co-clustering algorithm in HFCR.…”

“…As an additional note, HFCR is an enhanced version of one existing dual-partitioning-based algorithm, called Fuzzy Co-clustering with the Ruspini's condition (FCR), that we proposed earlier [16]. While the two algorithms are related, the heuristic formulation of HFCR allows it to be more robust to noises than FCR.…”