Um estudo comparativo entre técnicas de aprendizado de máquina para classificação de intenções, em um contexto de diálogos no e-commerce de roupa

Abstract

Over the past years, it has been noticeable how conversational agents, such as chatbots and virtual assistants, have been increasingly integrated into human life. By explicitly analyzing the e-commerce sector, it is possible to see how this type of technology is incorporated, bringing benefits to the final user. Over time these technologies have evolved and thanks to artificial intelligence, the process of generating value has been refined. Intent classification is one of the most important steps in building a modern conversational agent. Through the intent classifier, the chatbot can automatically identify and classify each statement entered by the user into intents, generating valuable knowledge. This work proposes an intention classifier approach for utterances in a dialogue in the context of fashion e-commerce. To achieve the objective, a comparative study was carried out between several machine learning models commonly used for this purpose. Results indicated that the SVC machine learning model is the one that demonstrates the best performance compared to other models in the classification of intentions for dialogues in e-commerce clothing.