Línguas naturais e máquinas artificiais: Aplicação de técnicas de mineração de texto para a classificação de sentenças judiciais brasileiras

This paper investigated the usage of artificial intelligence and text mining techniques for classification of court judgments and discussed potential alternative applications in formulation and evaluation of public policies. Besides, we built a survey of studies related to Jurimetry based on the specialized scientific literature and detailed the operationalization of the of textual data treatment, as well as basic concepts and methods of text mining. Finally, we performed an empirical analysis of classification of legal texts into four categories using real data from the Brazilian 2nd Federal Regional Court collected by IpeaJus, the database about the Brazilian Justice System from Ipea, discussing the results in light of various quantitative evaluation metrics and prospects for future developments in different contexts.