Persistent Community Search in Temporal Networks

Community search is a fundamental graph mining task. In applications such as analysis of communication networks, collaboration networks, and social networks, edges are typically associated with timestamps. Unfortunately, most previous studies focus mainly on identifying communities in a network without temporal information. In this paper, we study the problem of finding persistent communities in a temporal network, in which every edge is associated with a timestamp. Our goal is to identify the communities that are persistent over time. To this end, we propose a novel persistent community model called (theta ,tau)-persistent kappa-core to capture the persistence of a community. We prove that the problem of identifying the maximum (theta ,tau)-persistent kappa-core is NP-hard. To solve this problem, we first propose a near-linear temporal graph reduction algorithm to prune the original temporal graph substantially, without loss of accuracy. Then, in the reduced temporal graph, we present a novel branch and bound algorithm with several carefully-designed pruning rules to find the maximum (theta ,tau)-persistent kappa-cores efficiently. We conduct extensive experiments in several real-world temporal networks. The results demonstrate the efficiency, scalability, and effectiveness of the proposed solutions.