Uso de Machine Learning na inspeção de processos

Abstract

The ordinary procedural inspection is a procedure that occurs annually in the federal courts, with the objective of inspecting the progress of the processes in progress. Although there is a system called Instant to assist in the inspection, this action still happens in a very manual way. The automation of the inspection act is necessary to guarantee a continuous and independent inspection, without altering the normal functioning of the court. This study uses two different databases for input, and the same machine learning pipeline to train models capable of identifying the occurrence of inspection in processes. The results obtained in this approach came out as expected, due to the volume of data and structure of the process, which is a file composed of several types of documents. It highlights the importance and necessity of applying artificial intelligence to assist in the inspection process in comparison to the current inspection system.