Preprocessing data sets for association rules using community detection and clustering: a comparative study

Renan de PaduaExupério Léo Silva JuniorLaís Pessine do CarmoVeronica Oliveira de CarvalhoSolange Oliveira Rezende

Association rules are widely used to analyze correlations among itemson databases. One of the main drawbacks of association rules mining is that thealgorithms usually generate a large number of rules that are not interesting orare already known by the user. Finding new knowledge among the generatedrules makes the association rules exploration a new challenge. One possiblesolution is to raise the support and the confidence values, resulting in the generationof fewer rules. The problem with this approach is that as the support andconfidence values come closer to 100%, the generated rules tend to be formed bydominating items and to be more obvious. Some research has been carried outon the use of clustering algorithms to prepare the database, before extracting theassociation rules, so the grouped data can consider the items that appear onlyin a part of the database. However, even with the good results that clusteringmethods have shown, they only rely on similarity (or distance) measures, whichmakes the clustering limited. In this paper, we evaluated the use of communitydetection algorithms to preprocess databases for association rules. We comparedthe community detection algorithms with two clustering methods, aimingto analyze the generated novelty. The results have shown that community detectionalgorithms perform well regarding the novelty of the generated patterns.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: