Edge-Attributed Community Search for Large Graphs

Community search over large graphs has attracted huge attention in recent years. To achieve better cohesiveness and efficiency, new attributed variants of community search algorithms were proposed in the last two years. The attributes involved in those works are associated with nodes and thus they are used to represent the characteristics of the nodes. However, the characteristics of users in large social networks are usually retrieved from their interactions with each other. Since the relationships and interactions are modeled as edges in the graph, we propose to consider edge attributes and design a community search algorithm accordingly. In this paper, we present an edge-attributed community search algorithm which answers each community search query in linear time. The algorithm is tested on the DBLP dataset to show its efficiency and effectives, and we also compare it with the previous vertex-attributed community search algorithms to show its utility.