Privacy Preserving Data Mining On Relational Streaming Data
Ashish Mane, Pankaj Agarkar, Deepali Patil · 2015
Today every sector, whether it is business, educational, Medical, Military etc, required to store and handle large amount of information. Organization publishes the required data for data miner. Privacy of this information has become an important issue. This paper focuses on the privacy of sensitive attribute data. Here relational streaming data is used. Previous data mining applications were used to prefer the single level trust approach, in which data owner trust all data miner at single trust level, and only single perturbed copy was generated. In this paper Multilevel Trust approach is used, in which multiple perturbed copies of original sensitive attribute data is generated. Additive data perturbation approach is used to add the noise to sensitive attributes and group generation algorithm is used to generate the perturbed copies of original sensitive attribute data. In addition, when the records of database are updated, immediately new perturbed copies for that updated records are generated.