Application of natural language processing techniques for the analysis of the observance of selected articles of the Labor Reform (Law 13.467/2017)
Abstract
This article presents the technical approach based on artificial intelligence, used in a project to analyse the observance of the Labor Reform in judicial proceedings of the first instance of the Labor Tribunals of the 2nd, 3rd and 15th Regions, with decisions from 11/11/2017 to 30/05/2018. Its applicability was examined within the following themes : (a) Hours in itinere – Art. 58; (b) Outsourcing – Art 4º-A; Monetary correction – Art. 879; (d) Union contribution – Art. 579; (e) Intra-journey interval – Art. 71; and (f) Negotiated over the legislated – Art. 611-A. The study successfully employed natural language processing techniques to classify the decisions that refer to the labor reform and the framing of the merits of the decision in "do not apply" and "admits application". The classifications were constructed following an approach based on semantic dictionaries for each theme, with the determination of classes, guided by the technique of Feature Engineering; and the construction of thermo-document matrices from TFDCF and TFIDF techniques.Downloads
Published
2019-04-22
Issue
Section
Artigos (Papers)