An Efficient Multiresolution Clustering for Motif Discovery in Complex Networks

Motif discovery and network clustering in complex networks have received a lot of attention in recent years, also they are still challenging tasks in bioinformatics, big data analytics and data mining applications. Motif discovery in big data networks has a lot of important applications in different domains such as engineering, bioinformatics, cheminformatics, genomics, sociology and ecology for revealing hidden frequent structures, functional building blocks, or knowledge discovery. In this paper, a motif localization method based on a novel clustering algorithm in complex networks is presented. In our method, for each complex network, a novel structure so-called Augmented Multiresolution Network (AMN) is generated, then it is adaptively partitioned into several clusters and their corresponding subnets. Then top ranked subnets are chosen to discover network motifs. We show that the proposed method provides an efficient solution for clustering and motif discovery; It speeds up current motif discovery algorithms by pruning non-promising regions of complex networks. Experimental results show our algorithm efficiently deals with complex networks representing large datasets with high-dimensionality such as big scientific data. Our method also provides motivations for future studies in big data and complex networks.