MultiAspectForensics: mining large heterogeneous networks using tensor

Abstract

Modern applications such as web knowledge bases, network traffic monitoring and online social networks involve an unprecedented amount of 'heterogeneous' network data, with rich types of interactions among nodes. How can we find patterns and anomalies for heterogeneous networks with millions of edges that have high dimensional attributes, in a scalable way? We introduce MultiAspectForensics, a novel tool to automatically detect and visualise bursts of specific sub-graph patterns within a local community of nodes as anomalies in a heterogeneous network, leveraging scalable tensor analysis methods. One such pattern consists of a set of vertices that form a dense bipartite graph, whose edges share exactly the same set of attributes. We present empirical results of the proposed method on three datasets from distinct application domains, and discuss insights derived from these patterns discovered. Moreover, we empirically show that our algorithm can be feasibly applied to higher dimensional datasets.