Fast Uncovering of Graph Communities on a Chip: Toward Scalable Community Detection on Multicore and Manycore Platforms

Abstract

Graph representations are pervasive in scientific and social computing.They serve as vital tools to model the interplay among differentinteracting entities.In this paper, we visit the problem of community detection, which isone of the most widely used graph operations toward scientific discovery.Community detection refers to the process of identifying tightlyknitsubgroups of vertices in a large graph. These sub-groups or communitiesrepresent vertices that are tied together through commonstructure or function. Identification of communities could help in understandingthe modular organization of complex networks. However,owing to large data sizes and high computational costs, performingcommunity detection at scale has become increasingly challenging.Here, we present a detailed review and analysis of some of the leadingcomputational methods and implementations developed for executingcommunity detection on modern day multicore and manycorearchitectures. Our goals are to: a define the problem of community detectionand highlight its scientific significance; b relate to challengesin parallelizing the operation on modern day architectures; c providea detailed report and logical organization of the approaches that havebeen designed for various architectures; and d finally, provide insightsinto the strengths and suitability of different architectures for communitydetection, and a preview into the future trends of the area. It is ourhope that this detailed treatment of community detection on parallelarchitectures can serve as an exemplar study for extending the applicationof modern day multicore and manycore architectures to othercomplex graph applications.