Simulação e avaliação de comportamentos em sistemas multi-agentes baseados em modelos de reputação e interação

Abstract

This work was developed in the context of human behavioral simulation and group interaction study. Our main goal was to propose, to study, to simulate and to evaluate a cooperative multi-agent model based on Interaction and Reputation models. This Master Thesis aims to study and discuss about: (1) questions related to multi-agent systems concerning reputation, agents grouping and relationships between agents (cooperation, betray, trust); (2) the adoption of the simulated IPD (Iterated Prisoner’s Dilemma – a classical problem studied in Game Theory) as a tool for agents’ behavior analysis, which act individually or in groups; (3) a behavioral agent model that includes attributes and properties, as for example, reasoning about reputation, personality and ability to influence (or be influenced) by other agents; (4) the agent x agent interaction, considering different agent behaviors, analyzing the impact of their behaviors when faced to the Iterated Prisoner’s Dilemma problem. In this work we propose an experime