AgentDevLaw: um meta-modelo e um middleware para a integração de ontologias legais e sistemas multiagentes

Abstract

With multi-agent systems (MAS), social simulations are developed with the aim of understanding and developing new ideas about society. These simulations can occur through various themes, in the area of games, traffic, response to claims, about policies, among others. These surveys seek to understand how social exchanges are built and how agents can interact to solve various problems. Agents may need to know more about their actions and the environment in which they operate (what to do, how to do it, and what rules are in place). In this sense, agents use ontologies to recognize and process a variety of knowledge. Ontologies are descriptions of the world as human beings relate things. In the area of Computer Science, there are efforts to model human knowledge and distribute this information in ontology so that they can be interpreted in a computational environment. At some point, it is necessary to regulate the actions of these agents, restricting their behavior. Legal ontologies are a specific type to deal with modeling of norms, regulations, functions, and roles, coming from the legal area. Thus, this work develops a model of legal ontology and middleware that can assist the various platforms of agents in recognizing and interpreting the legal information of these ontologies. The knowledge base of this ontology is in Brazilian law and how it can be recognized by multi-agent systems. In the end, tests of applications of this proposal are carried out with simulations of regulation of actions and the creation of new laws by the agents (proposal of new laws by the Chamber and Senate). The results of the tests demonstrate the applicability and how the approach helps in the operation and understanding of the laws by the agents.