Multi-level and multi-key trust in PPDM

Privacy in data for data mining is ensured by many methods since decades. Data Transformation in the KDD process ensures transforming the data into cryptic codes and some abbreviated forms, yet the details of the data are guessable to the data miners. Perturbation theory comprises mathematical methods that are used to find an approximate solution to a problem which cannot be solved exactly, by starting from the exact solution of a related problem. Perturbation function causes a minor or major change in the result of the problem-solution scenario to get the expected yield mathematically. The concern over privacy of personal and sensitive information has led to the implementation of several techniques for hiding, obfuscating and encrypting sensitive information in databases. The requirement of preserving privacy as well as the usability of sensitive data has led to development of nearest neighborhood techniques. In this work we propose a method that expands the scope of perturbation in PPDM as multilevel and multikey trust in privacy preserving data mining. An analogical approach with measuring the identification attacks, diversity attacks and the problem is addresses the challenge by properly correlating perturbation across copies of different trust levels and keys that are pertaining to the sub domain contexts of the databases. Our proposed framework is architecturally robust and defends the attacks to achieve the privacy goal. Our framework supports data providers to deliver different forms of data with different privacy levels based on the market demand. Data owners can relax with the framework provided in this paper.