An argumentation model for deliberation in heterogeneous multiagent systems

• Autor(a) principal: Daniel José Marques Nora
• Data de Publicação: 2015
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: https://repositorio-aberto.up.pt/handle/10216/83520
• topic: Engenharia electrotécnica, electrónica e informática; Electrical engineering, Electronic engineering, Information engineering

The perception of the society about the negative impact of transportation on quality of life and on the environment has greatly increased over the last few decades. To be capable of satisfying the transportation requirements of a region, planners must design efficient transit facilities while respecting environmental, social, economic and legal factors. This task is highly complex and error-prone due to the overwhelming number of issues to take into account, pertaining to widely different areas of expertise. Ideally, planners should be assisted by good decision support systems to efficiently discover the best solutions in each situation. This principle applies to many other complex design problems. The aim of this thesis project is to develop an argumentation model for deliberation in heterogeneous multiagent systems. The model shall be generic enough to be easily adaptable to different application contexts, facilitating the construction of decision support systems based on argumentative agents. Such agents specialize in different areas of knowledge, and collaborate in search of solutions that optimize the performance of the system while respecting the different quality parameters valued by each other. This approach should be particularly useful when the problems to solve are too complex to employ traditional optimization algorithms. Additionally, users will benefit from argumentation-based justifications for the system's suggestions.Several key concepts pertaining to decision-making processes and deliberation dialogues are used as reference for the development of the model, which is split into three major components: an abstract argumentation framework for deliberation; a set of criteria for choosing decisions for individuals and groups, based on accepted arguments, priorities between goals and concepts from social choice theory; and a communication protocol enabling agents to interact and exploit the theory in connection with the previous components, allowing integration of innovating techniques, such as simulation, in order to take full advantage of the agents' analytical abilities. The application domain used for demonstrating the model's capabilities is that of urban transportation networks.