MSDMT

This repo is the TF2.0 implementation of Multi-source Data Multi-task Learning for Profiling Players in Online Games (MSDMT) [PDF].

MSDMT is a novel Multi-source Data Multi-task Learning approach for profiling players with both player churn and payment prediction in online games. On the one hand, MSDMT considers that heterogeneous multi-source data, including player portrait tabular data, behavior sequence sequential data, and social network graph data, can complement each other for a better understanding of each player. On the other hand, MSDMT considers the significant correlation between the player churn and payment that can interact and complement each other.

Folders

• data/: data of MSDMT (randomly generated sample data to show the data format, not the real data).
  • sample_data_player_portrait.csv: the sample data for player portrait.
  • sample_data_behavior_sequence.csv: the sample data for behavior sequence.
  • sample_data_social_network.csv: the sample data for social network.
  • sample_data_label.csv: the sample data for label, where label1 is churn label (binary classification) and label2 is payment label (regression).
• src/: implementations of MSDMT.
  • model.py: the code for model.
  • main.py: the code for pipeline.

Requirements

The code has been tested running under Python 3.5.2, with the following packages installed (along with their dependencies):

• tensorflow == 2.1.0
• spektral ==1.0.3
• numpy == 1.18.2
• pandas == 0.23.4
• sklearn == 0.19.1

Training

$ cd src
$ python main.py 

Cite

Please cite our paper if you use this code in your own work:

@inproceedings{zhao2020multi,
  title={Multi-source Data Multi-task Learning for Profiling Players in Online Games},
  author={Zhao, Shiwei and Wu, Runze and Tao, Jianrong and Qu, Manhu and Li, Hao and Fan, Changjie},
  booktitle={2020 IEEE Conference on Games (CoG)},
  pages={104--111},
  year={2020},
  organization={IEEE}
}