Inline Algorithm

In this chapter, we present in detail the first algorithm that has been proposed which does not rely on any heuristics to secure the sensitive knowledge derived by association rule mining. Similarly to Menon’s algorithm (covered in Chapter 13), the inline algorithm of [23] aims to hide the sensitive frequent itemsets of the original database

$$\mathcal{D_O}$$

that can lead to the production of sensitive association rules. As a first step, we will introduce the notion of distance between two databases along with a measure for quantifying it. The quantification of distance provides us with important knowledge regarding the minimum data modification that is necessary to be induced to the original database to facilitate the hiding of the sensitive itemsets, while minimally affecting the nonsensitive ones. By trying to minimize the distance between the original database and its sanitized counterpart, the inline algorithm formulates the hiding process as an optimization problem in which distance is the optimization criterion. Specifically, the hiding process is modeled as a CSP which is subsequently solved by using a technique called Binary Integer Programming (BIP). The attained solution is such that the distance measure is minimized.