We describe a web-based system for defining and running \tac experiments, and collecting results. It is a research tool, intended to support experimentation and evaluation of alternate agent configurations. With it, we are able to set up sets of experiments, each of which runs multiple games with a fixed set of competing agents. Game and agent logs are gathered together for later analysis. Experiment sets can be queued, and given enough hardware, multiple games can be run in parallel.
|Title of host publication
|Workshop: Trading Agent Design and Analysis (TADA) at Twenty-First International Joint Conference on Artificial Intelligence (IJCAI 2009)
|Number of pages
|Published - 2009