An optimized privacy model based on K anonymity for classification utility privacy preserving data mining

Abstract

The tremendous growth in Information Technology has resulted in large volumes of data which necessitate sophisticated analytics procedures for knowledge extraction and decision making Business Intelligence Decision Support Systems and Predictive Analytics for big data are the growing state of the art technologies However the underlying procedures to extract meaningful information are the data mining functionalities Privacy has become an important concern in the process of data mining Presence of subject specific information in the medical records or census data may result in the violation of individuals privacy Government organizations and Healthcare foundations often collect and publish micro data about individuals adhering to certain privacy policies Analysis of such data yield useful inferences to researchers and medical practitioners. Recently Evidence Based Medicine EBM is becoming a standard for clinical decision making EBM integrates the evidences and research de facto standards to clinical expertise in order to make diagnostic decisions for the patients Data sharing and publishing a detailed view of patients are the important agenda in EBM clinical decisions There are situations where the data analysis may not be used for such legitimate objectives An adversary may try to capture the identity of the data owners whose information is present in the published records The area of Privacy Preserving Data Mining PPDM aims at publishing data without compromising the individuals privacy newline