Valve's Team Fortress 2 (TF2) is a vastly popular online first-person shooter (FPS) with a fanbase and support that stems back a decade. It's rare that a FPS continues to carry such a strong community. However, it's fair to say that TF2 has continued to evolve as a platform, with significant changes to this weird and eclectic shooter since its original launch back in 2007. But it's not the game itself I'm interested in today -- given that I typically talk about the AI that works behind the scenes. Rather, I'm going to talk about at its community.

D. Fox Harrell (fox@csail.mit.edu) is Professor of Digital Media in both the Comparative Media Studies Program and the Computer Science and Artificial Intelligence Laboratory at the Massachusetts Institute of Technology, Cambridge MA, and the founder and director of the Imagination, Computation, and Expression Laboratory. Chong-U Lim (culim@csail.mit.edu) recently completed his Ph.D. in electrical engineering and computer science from the Computer Science and Artificial Intelligence Laboratory at the Massachusetts Institute of Technology, Cambridge MA, where he was a member of the Imagination, Computation, and Expression Laboratory.

D. Fox Harrell, associate professor of digital media with appointments in the MIT Comparative Media Studies/Writing program and in the MIT Computer Science and Artificial intelligence Laboratory (CSAIL), has recently been awarded several grants to advance his research at the intersection of the social sciences and digital technology. These grants, funded by the National Science Foundation (NSF), the MIT CSAIL Qatar Computing Research Institute (QCRI) Alliance, and the MIT Center for Art, Science, and Technology (CAST), together amount to $1.35 million in support for Harrell's groundbreaking interdisciplinary research. Harrell's new set of complementary initiatives builds upon his NSF CAREER Grant research project, "Computing for Advanced Identity Representation," to delve more deeply into the dynamic relationship between virtual avatars and personal identities. He was able to push this work in innovative new directions while spending the 2014-15 year as a fellow at Stanford University in the Center for Advanced Study in the Behavioral Sciences (with the support of the Lenore Annenberg and Wallis Annenberg Fellowship in Communication). "For five years I have been researching this intersection between human experiences and our identities as implemented across digital technologies such as video games and social media," Harrell says.

Videogame avatars provide an expressive avenue for players to represent themselves virtually. Research has shown that these avatars, while virtual, can reveal aspects of players' identities, along with physical, social, and cultural values of the real-world. In this paper, we present an approach for modeling player values through their avatars using artificial intelligence (AI) clustering techniques. In a study with 191 participants who created avatars using our system, we provide a thorough comparison of the techniques across numerical, textual, and visual data. Our findings showed that these data structures can effectively reveal players' values and preferences, such as conforming to stereotypes of character roles using statistical attributes, modeling nuances in text descriptions of avatars, and identifying "best-example" (prototypical) avatar appearances that players can be quantitatively shown to conform to. Our findings suggest that AI clustering approaches can be used to model players to yield insight into implicitly held values in a data-driven manner through virtual avatars.