Database Reference
In-Depth Information
Chapter 3
Classes of Association Rule Hiding
Methodologies
In this chapter, we present a taxonomy of frequent itemset and association rule hid-
ing algorithms after having reviewed a large collection of independent works in this
research area. The chapter is organized as follows. Section 3.1 presents a set of four
orthogonal dimensions that we used to classify the existing methodologies by tak-
ing into consideration a number of parameters related to their workings. Following
that, Section 3.2 straightens out the three principal classes of association rule hid-
ing methodologies that have been proposed over the years and discusses the main
properties of each class of approaches.
Fig. 3.1: A taxonomy of association rule hiding approaches along four dimensions.
3.1 Classification Dimensions
Figure 3.1
presents a set of four orthogonal dimensions based on which we classified
the existing association rule hiding algorithms. As a first dimension, we consider
whether the hiding algorithm uses the support or the confidence of the rule to driving
the hiding process. In this way, we separate the hiding algorithms into support-
based and confidence-based. Support-based algorithms hide a sensitive association
rule by decreasing the support of either the rule antecedent or the rule consequent,
or by lowering the support of the rule's generating itemset up to the point that the
support of the rule drops below the minimum support threshold. Confidence-based
