Considering Re-occurring Features in Associative Classifiers

The classification problem is one of the most common tasks in Data Mining and Machine Learning. Given its vast applicability in many real domains, supervised classification has been addressed and extensively studied. There are numerous different classification methods; among the many we can cite associative classifiers. This newly suggested model uses association rule mining to generate classification rules associating observed features with class labels. Given the binary nature of association rules, these classification models do not take into account repetition of features when categorizing. Repetitions of features are often good indicators and discriminators of classes, in particular for text or other multimedia. In this paper, we enhance the idea of associative classifiers with associations with re-occurring items and show that this mixture produces a good model for classification when repetition of observed features is relevant in the data mining application at hand.