Classificação automática de falhas em arquitetura orientada a serviços

• Ano de defesa: 2017
• Autor(a) principal: Felix, Kleber Gonçalves
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Tecnológica Federal do Paraná; Curitiba; Brasil; Programa de Pós-Graduação em Computação Aplicada; UTFPR
• Palavras-chave em Português: Sistemas operacionais distribuídos (Computadores); Localização de falhas (Engenharia); Controle automático; Aprendizado do computador; Arquitetura orientada a serviços (Computador); Algorítmos; Métodos de simulação; Computação; Distributed operating systems (Computers); Fault location (Engineering); Automatic control; Machine learning; Service-oriented architecture (Computer science); Algorithms; Simulation methods; Computer science; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; Ciência da Computação
• Link de acesso: http://repositorio.utfpr.edu.br/jspui/handle/1/2822

A distributed architecture is composed of many systems that exchange messages between each other. Faults in the integration of these systems may occur and they required a detailed investigation of support professionals to identifying the root cause of the problem. The manual process to identify causes of failure is difficult and time-consuming. Significant efficiency gains can be achieved by automating the faults classification process. This work presents a method to support the automated fault diagnostic process, automatically classifying faults generated in a Service Oriented Architecture (SOA). This method denominated SOAFaultControl, may be executed in a distributed architecture that adote SOA and an Enterprise Service Bus (ESB). Using machine learning techniques, was possible build a model to classify fault messages captured in a SOA environment, in pre-established classes. To achieve the objectives of this work it was necessary to test the following machine learning algorithms: Support Vector Machine, Naive Bayes, and AdaBoost. Results show that Support Vector Machine algorithm achieved better performance in the following metrics: precision, accuracy, recall, and F1.