Customer complaint classification using natural language processing: systematic literature review

• Autor: Flores Poma, José Luis
• Formato: artículo
• Fecha de Publicación: 2023
• Institución: Universidad Nacional Mayor de San Marcos
• Repositorio: Revistas - Universidad Nacional Mayor de San Marcos
• Lenguaje: español
• OAI Identifier: oai:ojs.csi.unmsm:article/27134
• Enlace del recurso: https://revistasinvestigacion.unmsm.edu.pe/index.php/rpcsis/article/view/27134
• Nivel de acceso: acceso abierto
• Materia: Natural language processing; machine learning; customer complaint; customer satisfaction; Procesamiento de lenguaje natural; queja de cliente; satisfacción de cliente

A dissatisfied customer with a product and/or service is motivated to express a complaint. Classifying complaints manually is a process that represents high costs in human and material resources. Artificial Intelligence (AI) allows the use of various algorithms to perform tasks that can simulate human intelligence, a branch of this is Natural Language Processing (NLP), its objective is that machines have the capacity to understand human language, allowing, for example, to classify and categorize data automatically. This article provides a systematic review of the literature addressing challenges in the classification of complaint texts, such as the lack of class balance, the presence of unlabeled data, and the interpretation of model results. Preprocessing techniques are explored, such as tokenization, stopword removal, and lemmatization, which influence model performance. Additionally, performance metrics such as precision, recall and F1-score are discussed. Current trends and future lines of research are shown. For this purpose, 24 articles published between 2018 and 2023 extracted from the Web of Science and Scopus databases were analyzed.