Uma abordagem ontológica para simulação de ação legal e consistência semântica aplicada à legislação brasileira

The SemanticWeb has emerged from the need to assign meaning to the contents made available on the web, so that, not only the human being but also the machines/ computers can process and understand such resources. It provides, therefore, a new infrastructure where programs and devices can exchange information in an intelligent and automated way, producing a “Web of Knowledge”. The Semantic Web has produced sophisticated services, with applicability in several areas. Notably, the rapid pace of organizational changes in public administration, the legal harmonization, the need to formalize and automate processes that involve a lawsuit, and the considerable increase in the legal corpus have attracted attention to the representation of Legal Knowledge, as well as the creation of systems that can reason in this field. This thesis, therefore, analyzes the nuances at the ontological level to formalize Brazilian Legal and Normative Knowledge, exploring an area known as Legal Computing. In the scope of Brazilian legislation, however, we can highlight semantic problems of linguistic as well as conceptual type. For the former case, the inherent ambiguity of natural language has led to the production of documents subject to multiple interpretations. In addition, the rhetorical and fancy language commonly used in law has led to the sub-specification of rules and concepts and, consequently, to inconsistencies, inaccurate modeling, and disagreements. This project aims, based on Semantic Web standards (the Description Logic formalism (DL) and the Ontology Web Language (OWL)), to define an ontological approach to model a portion of the Brazilian Criminal Law, in particular, the General Theory of Crime. Thus, it is possible to build systems based on legal-normative knowledge, capable of reasoning and supporting decision-making, mitigating potential syntactic and semantic anomalies. This thesis was still concerned with investigating the limits of classical logic (DL basis) in legal-normative knowledge, and how a non-monotonic logic can be used to resolve possible conflicts between laws. Thus, through the engineered Ontologies and a prototype developed, we highlight simulations of legal actions in real scenarios, extracting the different interpretations of the law, with their violations and penalties imposed. As contributions of this project, we highlight the theoretical and practical advance in the use of logically formalized ontologies for non-ambiguous and shared representation of legal knowledge, favoring transparency and greater effectiveness of the of laws. This thesis also helps to reduce the gap between technologies and legal systems.