O uso de inteligência artificial para melhorar a experiência dos clientes insatisfeitos em instituições financeiras

• Ano de defesa: 2024
• Autor(a) principal: Silva, Alexandre Pinto da
• Orientador(a): Balaniuk, Remis
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Católica de Brasília
• Programa de Pós-Graduação: Programa Stricto Sensu em Governança, Tecnologia e Inovação
• Departamento: Escola de Educação, Tecnologia e Comunicação
• País: Brasil
• Palavras-chave em Português: Instituição financeira; Reclamações; Inteligência artificial
• Palavras-chave em Inglês: Financial institution; Claims; Artificial intelligence
• Área do conhecimento CNPq: CNPQ::CIENCIAS SOCIAIS APLICADAS::CIENCIA DA INFORMACAO
• Link de acesso: https://bdtd.ucb.br:8443/jspui/handle/tede/3485

Customer dissatisfaction with a company’s services, when registered on the internet, can negatively influence the behavior of other consumers. However, these records, when used well by companies, can be useful in improving customer experience and satisfaction. Artificial Intelligence has demonstrated the potential to support companies in reversing customer dissatisfaction, improving company results. The objective of this research is to propose a system capable of supporting, guiding and advising the teams responsible for preparing responses, with a view to reversing the dissatisfaction of customers who register complaints against financial institutions. Using a mixed approach and using the CRISPDM process, experiments were carried out on textual bases of complaints collected on the ReclameAqui platform from the 15 most complained about financial institutions in the Bacen Ranking to predict customer dissatisfaction. Three experiments were carried out using Bertimbau, a pre-trained model in Portuguese. Experiment I achieved 83% accuracy in identifying the level of customer satisfaction. Experiment II was even better, achieving 93% accuracy. In Experiment III, complaints were divided into 5 groups. For groups 3 and 4, a predictive model similar to that of Experiment II was used, achieving an accuracy between 80% and 83%. With the results of the experiments, the systematic proposal for the area of customer complaints is to use Artificial Intelligence to predict the degree of customer satisfaction with the responses. This will help employees improve and personalize their responses, resulting in more efficient and satisfactory service.