B-A Cassell and MP Wellman

SIGMOD-14 Workshop on Data Science for Macro-Modeling, June 2014


There is a growing interest in computational game theory, the study of algorithms for analyzing multi-agent strategic interaction cast as games. Work in this field has primarily focused on algorithmic complexity of solving given game forms, with relatively little attention paid to concerns about managing data underlying game models. We propose a model of game-play data implementable by a relational database system, designed to scale to the vast quantities of observations needed to describe real-world strategic scenarios with sufficient fidelity. Our model enables performing common game-theoretic analysis in the database, using a flexible, compact representation of games.