Intention reconsideration like dichotomous choice: An argumentation-based approach

A key issue in the design of Belief-Desire-Intention (BDI) agents is that of finding an appropriate Intention Reconsideration (IR) strategy. IR presents a dichotomous choice to decide between deliberating or acting. This choice typically involves multiple criteria, and thus IR can be seen as a multi-criteria decision-making problem. Traditional approaches to IR define the strategy in the agent's design stage, which makes it impossible to modify it in execution time. This is clearly not a practical solution for agents operating in changing environments. That is why, in this work, we propose implementing the IR strategy with an argumentation-based mechanism that uses arguments accrual in Possibilistic Defeasible Logic Programming. This approach allows to change commitments to intentions depending on how the environment evolves, and takes into consideration multiple criteria when making a decision whether to reconsider or not.