A PPDM model using Bayesian Network for hiding sensitive XML Association Rules

Association Rule Mining (ARM) was introduced for the market basket analysis where items that are frequently appeared together per transaction are identified as rules. In such mining process, sensitivity issue of rules has never been addressed for more than a decade. Thus, research on guarding sensitivity in ARM should be attended to in priority by researchers, so that the risk of sensitive information disclosure can be avoided especially when the data sources are being shared. In this paper, we presented a Mode-based PPDM model via Bayesian Network (BN) which can reliably hide away sensitive rules in ARM. Such reliability was never studied nor reported in the literature of XML domain of PPDM. One useful advantage of PPDM model is its ability to unfasten a variety of directions that could be effectively used to overcome disclosure risk in XML Association Rules (XARs). Moreover, PPDM model is known to benefit businesses even in absolute competitive environment.