Second, four agent types are selected and implemented to compete namely, ABMP (A) [15, 16], Trade-Off (Y) , Bayesian (B) , Mining (M)  and the proposed Bayesian-Mining (BM) Agent.
To compare the performance of the Bayesian mining approach, the agents using opponent modelling were compared with agents using the ABMP, Trade-off, Bayesian and mining strategies.
In 3-issue experiments (Figure 1), after 5 sessions, when the agent B applies Bayesian--mining strategy to negotiate with the opponent A following the same strategy, Bayesian, Trade-off, Mining or ABMP, it gets the distances of the final agreement to the Pareto Frontier equal to 0.
When agent B follows the Trade-off strategy, it reaches better agreement when the opponent uses the Trade-off, then the Bayesian, and lastly ABMP.
For example, when Agent B follows Bayesian mining and plays against an opponent Agent A which follows ABMP on 5-issue data set (Figure 10), Agent B utilities outcomes are (0.
For example, agent A which follows ABMP (Figure 6) on 5-issue data set gets session rounds (75, 57, 43, 35) vs.