(FP)2-Growth. Minería de datos difusos en paralelo
Fecha
2021-09-10
Autores
Título de la revista
ISSN de la revista
Título del volumen
Editor
Jaén: Universidad de Jaén
Resumen
Actualmente, uno de los algoritmos mejor establecidos en el campo de la extracción de reglas de asociación es FP-
Growth, mejorando la eficiencia con respecto a anteriores propuestas. Pero uno de los problemas derivados de la minería
de datos está no sólo en la gran cantidad y variabilidad de la información, sino en la naturaleza de la misma. La teoría de
subconjuntos difusos permite tratar con información incierta o imprecisa. Una alternativa reciente es la representación
por niveles de restricción (RL), mediante la cual se descompone el concepto de conjunto difuso como una colección de
diferentes niveles, donde los elementos pertenecen con total certeza.
Para el presente trabajo se propone un estudio teórico-empírico para desarrollar una herramienta robusta y escalable que
nos permita extender el citado algoritmo FP-Growth para extracción de conocimiento en datos difusos, descomponiendo
la información en diferentes niveles de restricción, y procesando dichos niveles en paralelo.
Currently, one of the best established algorithms in the field of association rule extraction is FP-Growth, improving efficiency with respect to previous proposals. But one of the problems derived from data mining is not only in the great quantity and variability of the information, but in its nature. Fuzzy subsets theory allows us to deal with uncertain or imprecise information. A recent alternative is the representation by restriction levels (RL), by means of which the fuzzy set concept is decomposed as a collection of different levels, where the elements belong with full certainty. For the present work, a theoretical-empirical study is proposed to develop a robust and scalable tool that allows us to extend the aforementioned FP-Growth algorithm for knowledge extraction in fuzzy data, decomposing the information in different levels of restriction, and processing said levels in parallel.
Currently, one of the best established algorithms in the field of association rule extraction is FP-Growth, improving efficiency with respect to previous proposals. But one of the problems derived from data mining is not only in the great quantity and variability of the information, but in its nature. Fuzzy subsets theory allows us to deal with uncertain or imprecise information. A recent alternative is the representation by restriction levels (RL), by means of which the fuzzy set concept is decomposed as a collection of different levels, where the elements belong with full certainty. For the present work, a theoretical-empirical study is proposed to develop a robust and scalable tool that allows us to extend the aforementioned FP-Growth algorithm for knowledge extraction in fuzzy data, decomposing the information in different levels of restriction, and processing said levels in parallel.
Descripción
Palabras clave
Tratamiento inteligente de la información