B-A Cassell and MP Wellman

Francesca Giardini and Frédéric Amblard (eds.), Multi-Agent-Based Simulation XIII, Lecture Notes in Artificial Intelligence, vol. 7838, Springer, 2013.

originally presented at:

13th International Workshop on Multi-Agent Based Simulation, in conjunction with AAMAS, Valencia, June 2012.


Empirical game-theoretic analysis (EGTA) is a promising methodology for studying complex strategic scenarios through agent-based simulation. One challenge of utilizing this methodology is that it can require tremendous amounts of computation. Constructing the payoff matrix for a game of even moderate complexity entails significant data gathering and management concerns. We present EGTAOnline, an experiment management system that simplifies the application of the EGTA methodology to large games. We describe the architecture of EGTAOnline, explain why such a tool is practically important, and discuss avenues of research that are suggested through the use of EGTAOnline.