Discrimination Prevention and Privacy Preservation in Data Mining

Data mining is an increasingly more vital technology for extracting useful information hidden in large collections of data. Along with privacy, discrimination is a very vital problem when taking into account the legal and ethical aspects of data mining. Most of the human being  do not want to be discriminated just because of their, religion, gender, age, nationality and so on, when those attributes are used for building decisions about them especially like giving them a loan, job, insurance, etc. For this motivation, anti-discrimination methods including discrimination finding and avoidance have been introduced in data mining. It can be direct or indirect. When decisions are made depend on sensitive attributes then direct discrimination occurs. Indirect discrimination arises when decisions are based on non-sensitive attributes which are robustly correlated with biased sensitive ones. The proposed method gives samples of privacy preservation and potential discrimination in data mining, that privacy and discrimination risks should be handled collectively. We discover the association between privacy preserving data mining and discrimination avoidance in data mining to design methods capable of addressing both threats concurrently during the knowledge innovation process. This paper deals with the concepts for privacy preservation for direct and indirect discrimination avoidance in data mining.