Mineração de Processos em Microsserviços de uma Aplicação de Assistente Virtual

• Ano de defesa: 2023
• Autor(a) principal: Köerich, Fabio Gausmann
• Orientador(a): Rodrigues, Luiz Antônio
• Banca de defesa: Peres, Sarajane Marques, Assunção, Wesley Klewerton Guêz, Linz, Johannes Kepler University
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual do Oeste do Paraná; Cascavel
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação
• Departamento: Centro de Ciências Exatas e Tecnológicas
• País: Brasil
• Palavras-chave em Português: Chatbot; Descoberta de processo; Redes de petri; Pm4py; Inductive miner
• Palavras-chave em Inglês: Process mining; Microservices; Chatbot; Process discovery; Petri nets; Pm4py; Inductive miner
• Área do conhecimento CNPq: SISTEMAS DE COMPUTAÇÃO
• Link de acesso: https://tede.unioeste.br/handle/tede/6589

In current systems, there is a strong dependency between processes, and their proper functioning depends on the execution of each of them. In a microservices structure, despite the scalability and ease of implementation, given the large number of processing steps, there are difficulties in monitoring, challenges in managing and controlling operations, identifying work bottlenecks and failures, in addition to the challenge of representing their logical activities with the aim of capturing its behavior. The recording of events that occour in each microservice in the form of an event log facilitates the application of process mining, but its use is still recent and incipient. The objective of this work is to explore the potential use of process mining in a microservice-based Virtual Assistant application (chatbot) of a large financial institution. The study includes the steps of problem modeling, data extraction for event log generation, log handling, conversion to CSV format, and execution of process mining in terms of process model discovery and verification of compliance with two tools widely known in the area: a commercial one (Apromore), with an academic license (which only limits the amount of data that can be analyzed in a single project) and an open source tool (PM4PY). The study was conducted based on data obtained during the actual execution of the business process associated with the virtual assistant in a production environment. The results show the benefits of process mining when applied to a business process implemented in a heterogeneous and highly distributed architecture.