Opti2I and ε-precis Methods Selection Algorithm

Abstract

Since the reference algorithm APRIORI [AGR97], other algorithms for optimizing the extraction of association rules have been developed. But no method is generally better than the others. This article deals with the optimization of closed itemsets in the context of highly correlated data. The work in this article responds to one of the perspectives of our article entitled "A new approach to optimizing the extraction of frequent 2-itemsets". In this previous article, we had obtained interesting optimization results from the 2-itemsets on a context of extraction of scattered data (weakly correlated data). The present article allowed us to obtain interesting results of the 2-itemsets on dense data (strongly correlated). Our approach was inspired by the research work of {Pas00, CB02, BBR03, CF14]. It has improved the extraction of a concise number of association rules by introducing a margin of error defined by the parameter  in the formula δ δ Ferm (S)-δ <ε (δ an integer, δ>0, Ferm (S) is the -Closure of the