Date Added: Jun 2012
Data distribution across trusted agents by a source is complicated as there is always the danger of misappropriation by one or more number of users (agents). An alteration to this sensitive data using Watermarks is of no use. In case of a leak there are two possible situations. First the sensitive data is dormant in some individual's device. Second it is most probably published somewhere on the Web. Nothing can be done about the former scenario. The authors propose data allocation strategies like injecting "Realistic but fake" data records to the original sensitive data at random locations such that these fake tuples can be used to identify the leaker among the agents in a latter scenario.