Mostrar registro simples

dc.contributor.advisorRamirez Pozo, Aurora Trinidad, 1959-pt_BR
dc.contributor.otherUniversidade Federal do Paraná. Setor de Ciências Exatas. Programa de Pós-Graduação em Informáticapt_BR
dc.creatorGuimaraes, Alaine Margaretept_BR
dc.date.accessioned2024-02-02T13:43:16Z
dc.date.available2024-02-02T13:43:16Z
dc.date.issued2000pt_BR
dc.identifier.urihttps://hdl.handle.net/1884/25091
dc.descriptionOrientadora : Aurora Trinidad Ramirez Pozopt_BR
dc.descriptionDissertaçao (mestrado) - Universidade Federal do Paraná, Setor de Ciencias Exatas, Programa de Pós-Graduação em Informáticapt_BR
dc.description.abstractResumo: Um dos grandes desafíos da inteligência artificial (IA) é a simulação do comportamento humano através da implementação de agentes deliberativos. Esses agentes se apresentam como entidades computacionais dotadas de certo grau de autonomia e que raciocinam em função da representação dos estados mentais. O Grupo de Pesquisas da UFPR (Universidade Federal do Paraná) voltado para a área de Agentes desenvolveu um Sistema de Matrícula Inteligente - SISMAT [MATOO], onde os agentes foram definidos de acordo com os estados mentais de crença, capacidade e compromisso. Observamos que alguns processos importantes, como o de planejamento, não foram implementados e que não existe um raciocínio claro sobre a interação do agente com seu meio. Foi então projetado um agente deliberativo que simule esses estados, incluindo o planejamento e tenha seu conhecimento representado de forma explícita, dividido em componentes específicos através dos quais possa haver controle sobre ele próprio, e sobre sua interação com o mundo e com outros agentes. O agente desenvolvido, denominado AGENTE-M, simula o comportamento de um matriculador humano realizando o processo de matrícula de aluno em uma universidade. Suas tarefas estão distribuídas em componentes específicos, através dos quais mantém um raciocínio e controle sobre seu próprio processo, sobre o agente usuário com o qual interage e sobre o mundo. O agente usuário representa o aluno que deseja se matricular, e o mundo corresponde à base de dados referente ao curso, disciplinas e histórico escolar. O projeto do AGENTE-M foi baseado na arquitetura BDI (Beliefs, Desires and Intentions) proposta pelo framework DESIRE (DEsign and Specification of Interactive REasoning components), e seu protótipo foi desenvolvido na linguagem de representação de conhecimento LOOM. Os resultados obtidos mostraram que a arquitetura BDI torna o agente mais poderoso em termos de raciocínio, que seu projeto em componentes específicos facilita sua compreensão e manutenção, e que a linguagem LOOM permitiu implementar o agente de acordo como foi projetado, com algumas restrições.pt_BR
dc.description.abstractAbstract: One of the greatest challenges of the Artificial Intelligence (AI) is the human behavior simulation across the intelligent agents implementations. The agents are like computational entities with some degree of autonomy that reason in function of the representation of the mental states. The Group of Researches of UFPR (Federal University of Paraná) went back to the area of Agents and developed a System of Intelligent Registration - SYSIR [MATOO], where the agents were defined in agreement with the mental states of beliefs, capabilities and commitments. We observed that some important process, as planning, were not implemented and that a clear reasoning doesn't half exist about the agent's interaction with its. We decided then to project a deliberative agent that simulates those states and have its represented knowledge in an explicit way, divided in specific components through which it can have control on itself, and about its interaction with the world and other agents. The developed agent, denominated AGENT-M, simulates the behavior of a human accomplishing the process of student's registration in an university. Its tasks are distributed in specific components, through which it maintains a reasoning and control on its own process, the agent user with which interage and the world. The agent user represents the student that wants to register, and the world corresponds to the base of data regarding the course, disciplines and school records. The project of the AGENT-M is based on the BDI (Beliefs, Desires and Intentions) architecture proposed by the framework DESIRE (Design and Specification of Interactive REasoning components), and its prototype was developed at the knowledge representation LOOM language. The obtained results showed that the BDI architecture turns the most powerful agent in terms of reasoning, its project in specific components facilitates its understanding and maintenance, and the language LOOM allowed to implement the agreement agent as it was projected with some restrictions.pt_BR
dc.format.extentx, 98f. : il.pt_BR
dc.format.mimetypeapplication/pdfpt_BR
dc.languagePortuguêspt_BR
dc.relationDisponível em formato digitalpt_BR
dc.subjectInteligencia artificial distribuidapt_BR
dc.subjectCiencia da Computaçãopt_BR
dc.titleAgente-M : um matriculador inteligentept_BR
dc.typeDissertaçãopt_BR


Arquivos deste item

Thumbnail

Este item aparece na(s) seguinte(s) coleção(s)

Mostrar registro simples