SELECTING, SORTING AND RANKING ASSOCIATION RULES WITH MULTIPLE CRITERIA USING DOMINANCE RELATION
A. Dahbi, S. Jabri, Y. Balouki, and T. Gadi
Datamining is the process of extracting interesting knowledge and information of patterns from large databases. Using association rules in datamaining is one of the most relevant tasks in modern society, which aim to discover interesting relationship and correlation among sets of items in large transactional databases. One of the main problems related to the discovery of these association (that a decision maker can face) is the huge number of association rules extracted. Hence, the knowledge post-processing phase becomes very challenging to rank and select the most interesting AR, Various interestingness measures have been proposed as a post processing phase. However, the abundance of these measures caused a new problem because there is no optimal measure and there is no measure which is better than others. To overcome this challenge we propose a new algorithm based on dominance relation aiming to find a good compromise without favoring or excluding any measures. Numerical experiments and comparison with other approaches are made on benchmark datasets and confirm a significant performance of the proposed approach.
Keywords: Data mining, Association rules, Interestingness measures, Multiple criteria, dominance relation.