The Team Intelligence in On-line Tactical Games project is
investigating methods for analyzing and building human-level team
intelligence using the testbed of Counter-Strike. We have collected
numerous logs of championship-caliber human game plays and are
developing methods for analyzing them. Accomplishments to date include
the first formal methods for describing openings of Counter-Strike
games at a team level and a Hidden-Markov model for predicting
positions of unobserved enemy players. This work is carried out in
collaboration with Michael Bowling and his group at the University of
Alberta. More information is found on the project page