Wie is de Mol?
Coen Jonker, Jelle Nauta and Zhe Sun

Conclusion

Our final result provides a streamlined, robust simulator of WIDM. The GUI offers good insights into the logical considerations that govern the agents. We have provided multiple types of agents, three types of contestants and two types of moles. For further research it would be nice to investigate how well certain strategies work, depending on the roles of other agents. The logic of our agents involves the observations during assignments, conversations they have with eachother, and the conclusions they can draw from an elimination, based on the conversations. We make the rather strong assumption that all agents are truthful in their suspicions of who the mole is (except the mole of course), since this creates an elegant way of interpreting conversations. Our original aim of including liars would greatly complicate the logic.

Logic and probability

This model shows that it is possible to combine probabilistic models and logical reasoning while creating gaming agents. As probabilities are necessary to deal with uncertainty and uncertainty cannot be avoided in the real world, a combination of logic and probabilities may contribute towards intelligent logical agents that can cope with the real world.