Publication: Aplicación de algoritmos de inteligencia artificial a la optimización de la calidad de requisitos mediante sugerencias automáticas de mejora
Loading...
Identifiers
Publication date
2022-10
Defense date
2023-03-07
Authors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
La especificación de requisitos es de vital importancia en la planificación de un proyecto, es aquí en donde se especifican los límites y las bases sobre las que se va a sustentar dicho proyecto.
Esta tesis se ha centrado en la ingeniería de requisitos. Siendo la creación de requisitos de calidad, la mejora de la eficiencia y la automatización de tareas los objetivos principales.
Para poder alcanzar nuestros objetivos, se cuenta con 1035 requisitos que han sido previamente clasificados dependiendo de su calidad y posteriormente descritos por 26 atributos. De esta manera, los datos obtenidos sirven como muestra para extrapolar los conocimientos hacia cualquier otra base de datos.
Una de las fortalezas de esta tesis es el alcance del proyecto, se ha diseñado un sistema capaz de adaptarse a cualquier base de requisitos. Independientemente de los clasificadores utilizados. Una barrera que se ha conseguido sobrepasar gracias a la utilización de los algoritmos genéticos.
Para ello, se ha creado un método que se resume en los siguientes pasos: Primero se clasificarán los requisitos mediante la extracción de métricas de calidad que serán tomadas como base por el clasificador binario. En segundo lugar, se tomarán todos aquellos requisitos clasificados como de mala calidad y se utilizarán algoritmos genéticos para proponer soluciones de cambio optimizados de acuerdo con los costes de esfuerzo estimados.
La principal conclusión que se puede extraer es que los algoritmos genéticos nos pueden ofrecer soluciones interesantes aplicables en ingeniería de requisitos. Obteniendo así un ahorro de costes, automatizando tareas y favoreciendo una planificación más sólida y eficiente en cualquier proyecto.
The specification of requirements is of vital importance in the planning of a project, it is here where the limits and the bases on which the project will be based are specified. This research project has focused on requirements engineering. The main objectives are the creation of quality requirements, the improvement of efficiency, and the automation of tasks. To achieve our objectives, there are 1,035 requirements that have been previously classified depending on their quality and subsequently described by 26 attributes. In this way, the data obtained serve as a sample to extrapolate the knowledge to any other database. One of the strengths of the thesis is the scope of the project, a system capable of adapting to any base of requirements has been designed. Regardless of the classifiers used. A barrier that has been overcome thanks to the use of genetic algorithms. To do this, a method has been created that is summarized in the following steps: First, the requirements will be classified by extracting quality metrics that will be taken as a basis by the binary classifier. Second, all those requirements classified as bad will be taken and genetic algorithms will be used to propose optimized change solutions according to the estimated effort costs. The main conclusion that can be drawn is that genetic algorithms offer us interesting solutions applicable to requirements engineering. Thus obtaining cost savings, automating tasks, and favoring more solid and efficient planning in any project. project.
The specification of requirements is of vital importance in the planning of a project, it is here where the limits and the bases on which the project will be based are specified. This research project has focused on requirements engineering. The main objectives are the creation of quality requirements, the improvement of efficiency, and the automation of tasks. To achieve our objectives, there are 1,035 requirements that have been previously classified depending on their quality and subsequently described by 26 attributes. In this way, the data obtained serve as a sample to extrapolate the knowledge to any other database. One of the strengths of the thesis is the scope of the project, a system capable of adapting to any base of requirements has been designed. Regardless of the classifiers used. A barrier that has been overcome thanks to the use of genetic algorithms. To do this, a method has been created that is summarized in the following steps: First, the requirements will be classified by extracting quality metrics that will be taken as a basis by the binary classifier. Second, all those requirements classified as bad will be taken and genetic algorithms will be used to propose optimized change solutions according to the estimated effort costs. The main conclusion that can be drawn is that genetic algorithms offer us interesting solutions applicable to requirements engineering. Thus obtaining cost savings, automating tasks, and favoring more solid and efficient planning in any project. project.
Description
Mención Internacional en el título de doctor
Incluye: Fe de erratas, pp. 107-110.
Incluye: Fe de erratas, pp. 107-110.
Keywords
Inteligencia artificial, Ingeniería del software, Ingeniería de requisitos, Desarrollo de software, Algoritmos genéticos, Calidad del software