Data mining is an important technology for extracting useful information from large collections of data in the database. Data mining techniques like classification rule mining and automated data collections have given the way to making automated decisions for loan granting or denial, personal selection etc. If the training data sets are biased with discriminatory or sensitive attributes like gender, age, religion, color etc. , discriminatory decisions may ensue. That cause potential privacy invasion and potential discrimination. Later one consists of unfairly treating people on the basis of their belonging to a specific group. The anti-discrimination techniques named discrimination discovery and prevention have been introduced in data mining to solve these problems. Discrimination is divided into two, Direct and Indirect and it tackles discrimination prevention in data mining and propose new techniques applicable for direct, indirect and both at the same time. It also describes how to clean training data sets and outsourced data sets in such a way that direct and/or indirect discriminatory decision rules are converted to legitimate (nondiscriminatory) classification rules and a number of papers mention measures of utility too. This survey paper is aimed at understand the existing discrimination prevention techniques and the utility measures discussed so far.

Data mining Discrimination Prevention Direct and Indirect Discrimination