Developing Social Identity Models of Players from Game Telemetry Data


  • Chong-U Lim Massachusetts Institute of Technology
  • D. Harrell Massachusetts Institute of Technology



player modeling, game telemetry


In this paper, we present an approach to modeling aspects of the identities of videogame players by data mining game telemetry information on in-game player performance and customization preferences. Our model demonstrates that such data can be used to reveal aspects of the identities players express by their social networking profile information. We tested our model on players of the multiplayer first-person shooter videogame Team Fortress 2. It was able to significantly explain the variances of the players' number of friends (35.1%), number of uploaded screenshots (49.6%), and number of uploaded videos (39.2%) of their profiles on the gaming social network Steam. Our results revealed several findings, such as criteria indicating how players customized avatars differently according to notions of aesthetics and practicality, and how these notions contributed to predicting their number of friends on their social networking profiles.Responses evaluated from a conducted survey reaffirmed several of these findings.




How to Cite

Lim, C.-U., & Harrell, D. (2021). Developing Social Identity Models of Players from Game Telemetry Data. Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, 10(1), 125-131.