Elaboration of implicative graph according to measure MGK
The implicative graph is one of the statistical tools very essential for the representation of the sequence of attribute groups. It allows the user to highlight association rules for these attributes. Several studies have already been developed, but we find that these works focus only on two measure: measure of Gras and Lermans. While only using these two measures are not sufficient to guarantee the quality of graphs implicative. In this work, we focus on the elaboration of the implicative graph of all association rules by using another quality measure MGK. Such a construction is exponential in the size of database, and mainly due to the paths implicative step. It is therefore necessary to define an efficient algorithm to automate the construction. Following this study, we propose a new algorithm that generates an implicative graph using this new quality measure MGK. We conducted experiments using one database to test the performance of our algorithm.
Keywords: Algorithm, Implicative graph, Database, Association rules, Quality measure MGK.
Download Full-Text
ABOUT THE AUTHORS
Bemarisika Parfait
First Author: Bemarisika Parfait, holder of Master 2 in Mathematics and Applications, University of Toulouse I, France (2010). Currently, PhD student in Mathematics and Computer Science, under supervision of Professor Totohasina André and Professor Jean Diatta.
Totohasina André
Second Author: Totohasina André, Professor at the University of Antsiranana, Madagascar. Specialist in Mathematics and Applications, Probability and Statistics in particular. Director of Doctoral formation Mathematics and Computer Science and research didactic, said University.
Bemarisika Parfait
First Author: Bemarisika Parfait, holder of Master 2 in Mathematics and Applications, University of Toulouse I, France (2010). Currently, PhD student in Mathematics and Computer Science, under supervision of Professor Totohasina André and Professor Jean Diatta.
Totohasina André
Second Author: Totohasina André, Professor at the University of Antsiranana, Madagascar. Specialist in Mathematics and Applications, Probability and Statistics in particular. Director of Doctoral formation Mathematics and Computer Science and research didactic, said University.
