Communities discovery in mobile call records

Given the users relational data of real world, how to get a clear vision of the network structure? A direct way is to identify the primary communities contained in and composed of it. Our goal is to bring forward a pragmatic solution not only discovering such concerned communities but also offering some inner hierarchy information. We propose a novel core vertices based method. Compared with former analogous algorithms, it overcomes the intrinsic shortcomings of this kind of solutions. Our method requires no strong assumption and no user-specific confined parameters. It can be applied to large scale of data set and a great deal of application situations. Experiments on real data show that it is effective and efficient to offer a comprehensible accuracy outcome.