Classificação automatizada de ligações de uma central de atendimento ao cliente utilizando processamento de linguagem natural

Abstract

The areas of customer service have gained importance with the increase of digital services and increasing customer demands. In this scenario, tools that help in this process become more important to ensure fast and assertive service. This work seeks to develop tools to direct customer service to the responsible sector through automated analysis of the customer’s verbal report, using various combinations of natural language processing models and machine learning techniques. To check the quality of the developed models, multiple metrics and a reference model are used. The reference model was chosen for its demonstrated effectiveness in several natural language processing problems. As a result, the models developed prove to be very effective in classifying calls, reaching 90% overall accuracy and surpassing the reference model.