Probabilistic Community and Role Model for Social Networks

Please kindly refer to the paper:

@inproceedings{han2015probabilistic,
  title={Probabilistic community and role model for social networks},
  author={Han, Yu and Tang, Jie},
  booktitle={Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining},
  pages={407--416},
  year={2015},
  organization={ACM}
}

There are three input files, i.e., an attributes file, a topology file and an action file, containing the information of nodes' attributes, edges and nodes' action respectively. The program uses CRM model to fit the input data and learn the parameters of the model, outputting the parameters of the model into a file. You can use the parameters of CRM to carry out many tasks. The format of the input files are as follows.

The attributes file contains a matrix, where the number of the rows is the number of the nodes, and the number of the columns is the number of attributes of each node. The element located at the m-th row and n-th column represent the value of the n-th attributes of the m-th node. The ID of the nodes and the ID of the attributes all start with 0. The format of the attribute data is as:

0.2267 0.5556 0.9821 …

0.6678 0.2216 0.7789 …

0.1211 0.2897 0.6712 …

…

Each line of the topology file represents an edge, and contains two numbers which are the IDs of the two endpoints. The format of the topology data is as:

0 5

3 7

8 9

…

Each line of the action file represents an act record, and contains three elements which represent the performer of the act, the target of the act, and the ID of the act respectively. The ID of the act starts with 0. The format of the attribute data is as:

0 1 0

1 2 0

1 2 1

…

The code of the CRM program can be downloaded here. If you use this code in your research, please kindly cite the paper listed in the part of introduction.