Weighted Association Rule Mining Over Unweighted Databases Using Inter-Item Link Based Automated Weighting Scheme

Abstract

Weighted association rule mining is an effective approach in discovering hidden relationships among the important items in a transactional database. Weight of an item reflects its importance in the database. However, most of the traditional methods are suitable for weighted item transaction databases (WITDs) where the item weights are available. In case of the unweighted item transaction databases (UWITDs), these methods remain ineffective. Item weights are not available in the UWITDs, and hence, the task of weight assignment has become one of the prime issues in this respect. This paper presents an automated weight assignment scheme for the items in an UWITD using the inter-item links. Unlike the existing approaches, the proposed scheme considers the indirect links in addition to the direct links among the items. Indirect links adjust the weights of the items, which in later help in mining large itemsets with low supports. We propose a link-based weighted association rule mining approach over the UWITD. The proposed approach includes two new objective measures such as linkage weighted support and linkage weighted confidence for mining the frequent weighted itemsets (FWIs) and the weighted association rules (WARs), respectively. The comprehensive experiments on both of the synthetic and real-world datasets show the effectiveness of the proposed approach in terms of number of FWIs and WARs, runtime, memory usage, weight distribution, scalability and dissociation.