Um conjunto de abordagens para a geração da matriz de rastreabilidade de requisitos com suporte de técnicas de inteligência computacional

• Ano de defesa: 2014
• Autor(a) principal: Ditomaso, André
• Orientador(a): Fabbri, Sandra Camargo Pinto Ferraz
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação - PPGCC
• Palavras-chave em Português: Rastreabilidade de requisitos; Matriz de rastreabilidade de requisitos; Engenharia de software experimental
• Palavras-chave em Inglês: Requirements traceability; Requirements traceability matrix; Experimental software engineering
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::METODOLOGIA E TECNICAS DA COMPUTACAO
• Link de acesso: https://repositorio.ufscar.br/handle/ufscar/11782

Context: Problems in requirements management is considered one of the causes for system failures. One of the activities that helps the requirements management, making it more effective, is the requirement traceability. One of the most important artefacts to determine and monitor the traceability of requirements is the Requirements Traceability Matrix. Nonetheless, establish and maintain the Requirements Traceability Matrix is an extremely laborious and error prone task. Objective: The objective of this thesis is to present a set of approaches that allow to generate the Requirements Traceability Matrix in an automated way, using computational intelligence techniques that allows the generation of more accurate links. Methodology: The proposed approaches explore functional requirements data and natural language processing solutions for determining the traceability matrix. From some experimental studies, the approaches were refined and combined with computational intelligence techniques to increase the accuracy of the traceability links. Results: Four approaches were proposed. The first approach considers that that the dependence of the requirements is related to the data manipulated by them. The second approach explores the use of natural language processing. The third approach combines the previous two approaches with fuzzy systems. Finally, the fourth approach uses artificial neural networks taking the data from the first two proposed approaches as input data. The experimental studies produced satisfactory results for the traceability links determined by the proposed approaches. Case studies were also conducted to evaluate the use of the approaches in the industry. Conclusions: The obtained results support the thesis defended in this research, demonstrating that the use of computational intelligence techniques improves accuracy in traceability links. links.