A Novel Enumeration Strategy of Maximal Bicliques from 3-Dimensional Symmetric Adjacency Matrix

Abstract

Several algorithms are available in the literature to find all maximal bicliques from an adjacency matrix. If these algorithms are applied directly on a symmetric adjacency matrix, all maximal bicliques will be generated twice. We propose a novel algorithm, S-Datapeeler, to enumerate all maximal bicliques only once in the context of 3-dimensional symmetric adjacency matrix i.e. zero duplicate patterns are generated. We have compared our results with DataPeeler in 3-D context. The DataPeeler algorithm generates all duplicate patterns, whereas the proposed algorithm completely eliminates all duplicate maximal bicliques. The proposed methodology results in 50% reduction in search space and thereby running time of S-Datapeeler is better for 3-D symmetric datasets.