Database Reference
In-Depth Information
usually occurs when hiding itemsets with a very high support and/or when using
a very low minimum support threshold. In most of these cases, approaches like
the ones presented in Section 16.4 become essential to allow for a solution of the
hiding problem. It is important however to mention that for as long as the given
hiding problem decomposes to a CSP that remains computationally manageable and
a sufficiently large safety margin is used, the hybrid approach is bound to identify a
solution that will bear the least amount of side-effects to the original dataset. Thus,
contrary to state-of-the-art approaches, the power of the hybrid hiding methodology
is that it guarantees the least amount of side-effects to an extended set of hiding
problems, when compared to the inline algorithm of [23].
