Argumentation Based Decision Making for Trust in Multi-Agent Systems

Agents in Multi-Agent Systems depend on assistance from others to attain their goals. Often, goals of agents conflict with each other, and agents can be unreliable or deceitful. Therefore, rational agents embedded in an open Multi-Agent System need to be equipped with algorithms to reason about trust. Currently, existing algorithms mainly focus on performance of these algorithms in terms of utility, but do not provide explanation of their actions. This might hinder acceptance of agent-based technologies in sensitive applications where users rely on their personal agents. In this thesis, we therefore propose a new approach to trust in MultiAgent Systems based on argumentation that aims to expose the rationale behind these decisions. Our solution features a clear separation of opponent modeling and decision making. It uses fuzzy logic to model behavior of opponents, and argumentation to translate these models into wellsupported decisions.