A Flow-based and Information-Theoretic Method for Community Detection from Signed Networks

Abstract

Detecting communities in complex networks accurately is a great challenge. Although many community detection (CD) algorithms have been proposed, most of them were designed primarily for networks containing only positive links, while many actual networks also feature negative links, such as friends and foes, can be represented as signed networks that contain both positive and negative links. In most community detection algorithms, like modularity maximization approach, always interested in how a system’s network structure was formed, but ignore how a network’s extant structure influences the system’s behavior. They pay no attention to the patterns of flow on the network. In this paper we extend a flow-based and information-theoretic method witch known as the map equation to incorporate negative links as well. To illustrate our method, we applied it to extended LFR-Benchmark and compare the experiments result with existing community detection algorithm for signed networks.