Discrimination Prevention with Privacy Preservation in Data Mining

Discrimination is a crucial issue while considering the legality and morality of data mining. People never like to be distinguished on basis of their caste, gender, nationality, etc. specifically while making decisions with consideration of this attributes, like offering them a job, property, loan, and so forth. Also when a data is released for analysis, there are chances that personal data may be misused for different reasons. Therefore, many companies and institutions are trying to bridge the gap between mining and sharing user data so that their services are improved and user privacy is ensured. We are trying to adapt an approach which will publish the discrimination free data such that probability of learning sensitive value of individual will be reduced. Concept of Slicing is used for preserving privacy. Slicing is better approach of privacy publishing in comparison with other approaches like generalization and bucketization, as it preserves correlation between attributes and it prevents attribute disclosure. The input to the system is discriminated dataset and output will be the sliced data that helps to preserve privacy.