One important question remains. How generic is your agent? That is, does it work as well on the party domain as on other domains? To verify this, have your agent negotiate against itself, and against at least three agents provided to you on at least three scenarios (for example, ItexvsCypress, Pie and Smart Grid). Report on the outcomes reached. Try to explain why the outcome is as it is. Is any of the outcomes a Nash solution? Is it possible to reach an efficient outcome, i.e. an outcome that lies on the Pareto Frontier?
One important question remains. How generic is your agent? That is, does it work as well on the party domain as on other domains? To verify this, have your agent negotiate against itself, and against at least three agents provided to you on at least three scenarios (for example, ItexvsCypress, Pie and Smart Grid). Report on the outcomes reached. Try to explain why the outcome is as it is. Is any of the outcomes a Nash solution? Is it possible to reach an efficient outcome, i.e. an outcome that lies on the Pareto Frontier?