Database Reference
In-Depth Information
this item is removed from the transaction. Finally, it selects to remove the item from
the transaction that will affect the higher number of sensitive and the least num-
ber of nonsensitive itemsets. The third approach, called Hybrid, is a combination of
the two previous algorithms; it employs the Aggregate approach to identify the sen-
sitive transactions and the Disaggregate approach to selectively delete items from
these transactions, until all the sensitive knowledge is appropriately concealed.
Wu, et al. [79] propose a sophisticated methodology that removes the assump-
tion of [20] regarding the disjoint relation among the items of the various sensitive
rules. Using set theory, the authors formalize a set of constraints related to the possi-
ble side-effects of the hiding process and allow item modifications to enforce these
constraints. However, the existing correlation among the rules can make impossible
the hiding of the sensitive knowledge without the violation of any constraints. For
this reason, the user is permitted to specify which of the constraints he/she considers
more significant and relax the rest. A drawback of the approach is the simultaneous
relaxation (without the users' consent) of the constraint regarding the hiding of all
the sensitive itemsets. To accommodate for rule hiding, the new scheme defines a
class of allowable modifications that are represented as templates and are selected
in a one-by-one fashion. A template contains the item to be modified, the applied
operation, the items to be preserved or removed from the transaction and coverage
information regarding the number of rules that are affected. Based on this informa-
tion the algorithm can select and apply only the templates that are considered as
beneficial, since they cause the least side-effects to the sanitized database.
Pontikakis, et al. [59] propose two distortion-based heuristics to selectively hide
the sensitive association rules. The proposed schemes use efficient data structures
for the representation of the association rules and effectively prioritize the selection
of transactions for sanitization. However, in both algorithms the proposed hiding
process may introduce a number of side-effects, either by generating rules which
were previously unknown, or by eliminating existing nonsensitive rules. The first
algorithm, called Priority-based Distortion Algorithm (PDA), reduces the confi-
dence of a sensitive association rule by reversing 1's to 0's in items belonging in the
rule's consequent. The second algorithm, called Weight-based Sorting Distortion
Algorithm (WDA), concentrates on the optimization of the hiding process in an at-
tempt to achieve the least side-effects and the minimum complexity. This is achieved
through the use of priority values assigned to transactions based on weights. Both
PDA and WDA are experimentally shown to produce hiding solutions of compa-
rable (or slightly better) quality than the ones produced by the algorithms of [64],
generally introducing few side-effects. However, both algorithms are computation-
ally demanding, with PDA requiring typically twice the time of the hiding method-
ologies in [64] to facilitate the hiding of the sensitive knowledge.
Wang & Jafari [76, 77] propose two data modification algorithms that aim at the
hiding of predictive association rules, i.e. rules containing the sensitive items on
their left hand side (rule antecedent). Both algorithms rely on the distortion of a
portion of the database transactions to lower the confidence of the sensitive associa-
tion rules. The first strategy, called ISL, decreases the confidence of a sensitive rule
by increasing the support of the itemset in its left hand side. The second approach,
