Toponym Based Community Search in Large Social Network

The geo-social community search problem, which aims to find a group of socially and geographically cohesive users in large social networks, has been well defined and investigated. However, instead of coordinates, there are more and more text locations be used to identify users, so the traditional coordinate based geo-social community search algorithm are no longer applicable, and it is necessary to design an algorithm based on the semantics of text locations to measure cohesive. In this paper, in order to overcome the shortcoming of traditional algorithm, we first propose an text location based geo-social community search algorithm by constructing a toponym graph, toponym graph is a kind of retrieval data structure which abstracts the real geographic topological relationship. In toponym graph, we quantify the semantic distance between two text locations and determine whether the two users are geographically close, and then find the socially and geographically cohesive community. After that, To solve the problem of high time complexity of the basic algorithm , we design an approximate algorithm to simplify the search process by setting the landmark nodes. And then we develop a series of pruning strategies to speed up the search. Finally, we verify the performance of our proposed algorithm and pruning strategies using some real world datasets.