Database Reference
In-Depth Information
Chapter 8
Summary
In the second part of the topic, we presented some state-of-the-art algorithms
that have been proposed for association rule hiding which belong to the class of
heuristic-based approaches. The heuristic class of approaches collects computation-
ally and memory efficient algorithms that operate in a series of steps by optimizing
certain subgoals to drive the hiding process. We partitioned the current heuristic
approaches into two main categories: distortion-based schemes, which operate by
alternating certain items in selected transactions from 1's to 0's (and vice versa), and
blocking-based schemes, which replace certain items in selected transactions with
unknowns, in order to facilitate association rule hiding. Each category of approaches
was further partitioned into support-based and confidence-based methodologies, de-
pending on whether the algorithm uses the support or the confidence of the rule to
drive the hiding process. A large amount of research has been conducted over the
past years, leading to several interesting heuristic methodologies being proposed for
association rule hiding. In Chapters 6 and 7 it was our intention to cover a selection
of the existing methodologies by considering the most representative ones for this
domain. As a final remark, we should note down that, as is also evident from the
number of presented works in each category, research on heuristic methodologies
for association rule hiding has mostly concentrated on the direction of distortion-
based approaches rather than blocking techniques. However, blocking techniques
are certainly more preferable than conventional distortion methodologies in several
real life scenarios and for this reason we feel that such approaches are expected
to attract more scientific interest in the years to come. Moreover, the combination
of distortion and blocking techniques is also a prominent research direction, which
may lead to solutions that further minimize the data loss that is introduced to the
original database to account for the hiding of the sensitive knowledge.
