To search, Click below search items.


All Published Papers Search Service


Dissemination Of Sensitive Data


K Bindu Susan, M.Raja Babu


Vol. 13  No. 7  pp. 67-71


Defending sensitive data dissemination from adversary's activities like Record, Attribute, Table linkages is an enforcing aspect for better prospects. Existing popular protective measures like k-anonymity and l-diversity perform better in achieving overall data utility maximization by reducing the information loss incurred in the anonymizing process. Unfortunately their strengths are confined to fixed schema data sets with low dimensionality. Earlier two novel anonymization methods such as approximate nearest-neighbor (NN) search using locality-sensitive hashing (LSH) and data transformation techniques like reduced band matrix , gray encoded sorting are used to parse high-dimensional spaces. The random projection feature of LSH is a computation overhead. So we propose to replace the former method with a variant of a k-d Tree (Spill tree) that uses an overlapping splitting area to find nearest-neighbor(NN). NN-search using Spill trees has significant performance boost with superior data utility and best execution time. We show experimentally by using both real data sets (from UCI and KDD repositories) and also synthetic data sets designed to exercise the algorithms in various ways.


Anonymization ,k-anonymity,l-diversity,spill trees , Band matrix.