An Efficient Method For Discrimination Prevention Using Differentiated Virtual Passwords And Secret Little Functions

In classification, discrimination is a type of treatment that includes denying the membership in one group opportunities that are available in another group. Discrimination based on age, religion, gender, caste, disability, employment, language, race and nationality. In this technique, direct and indirect discrimination is prevented using rule protection and rule generalization methods and BIRCH (Balanced Iterative Reducing and Clustering using Hierarchies) is used to perform hierarchical clustering for huge data-sets. To enhance privacy, we propose a virtual password concept to secure user‘s passwords in banking process. We applied user-determined randomized linear generation functions to secure user‘s passwords based on the fact that a server has more information than any adversary does. We are evaluating metrics for proposed methods that impact on information loss and data quality in data mining.