Método para la agregación de rankings a partir de dos grupos con intereses contrapuestos
Archivos
Fecha
2015-06-26
Autores
Pérez Lugo, Lázaro Jesús
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas
Resumen
A partir de los rankings generados por dos grupos de expertos con intereses contrapuestos, se quiere realizar la agregación de los rankings de cada grupo buscando que los rankings resultantes de la agregación minimicen las diferencias respecto a los rankings del mismo grupo y a la vez maximizar dichas diferencias respecto a los del otro grupo. Para medir dichas diferencias se toma como medida de distancia entre rankings la Distancia de Kendall. Para buscar esta agregación se utiliza la metaheurística Algoritmo Genético. Se utiliza la plataforma evolutiva ECJ implementada en JAVA, de la cual se extiende para desarrollar el algoritmo. Para validar los resultados se realiza un estudio comparativo de los resultados alcanzados usando el algoritmo genético con los alcanzados mediante una búsqueda exhaustiva en la que todas las combinaciones de cromosomas posibles se tienen en cuenta; en particular se estudió los casos de rankings de 3, 6 y 10 candidatos. El algoritmo implementado sobre la plataforma evolutiva ECJ arroja resultados satisfactorios.
From the rankings generated by two groups of experts with opposite interests, it is needed to execute an aggregation of the rankings of each group looking for the resulting rankings to minimize the differences with respect to the ones of the same group and at the same time to maximize such differences with respect to the other group. For measuring those differences, the Kendall´s Distance is taken as the distance measure among rankings. In order to look for this aggregation, the Genetic Algorithm metaheuristic is used. The evolutionary platform ECJ implemented in JAVA is used, from which is extended to develop the algorithm. For validating the result, there is executed a comparative study of the accomplished results, using the genetic algorithm with the reached ones through an exhaustive search in which all the possible combinations of chromosomes are taken into account; rankings’cases from 3,6 and 10 candidates were study in particular. The algorithm implemented over the evolutionary platform ECJ yields satisfactory results.
From the rankings generated by two groups of experts with opposite interests, it is needed to execute an aggregation of the rankings of each group looking for the resulting rankings to minimize the differences with respect to the ones of the same group and at the same time to maximize such differences with respect to the other group. For measuring those differences, the Kendall´s Distance is taken as the distance measure among rankings. In order to look for this aggregation, the Genetic Algorithm metaheuristic is used. The evolutionary platform ECJ implemented in JAVA is used, from which is extended to develop the algorithm. For validating the result, there is executed a comparative study of the accomplished results, using the genetic algorithm with the reached ones through an exhaustive search in which all the possible combinations of chromosomes are taken into account; rankings’cases from 3,6 and 10 candidates were study in particular. The algorithm implemented over the evolutionary platform ECJ yields satisfactory results.
Descripción
Palabras clave
Método Computacional, Agregación de Rankings, Grupos de Expertos, Intereses Contrapuestos, Distancia de Kendall, Algoritmos Genéticos, Toma de Decisiones