Διασφάλιση μη-διάκρισης μέσω εξόρυξης δεδομένων
Discrimination - aware data mining

View/ Open
Abstract
The main aim of this dissertation was to combine the two different fields of Data Mining, the Privacy Preserving and the Discrimination-aware, for the benefit of humans. Privacy Preserving deals with the protection of humans’ sensitive information, included in databases that are ready to be published in order not to be leaked and protect the human privacy. Furthermore, Discrimination-aware is the field that deals with confronting discrimination, which is a common problem in classification. Discrimination is the phenomenon of unequal and unfair treatment of people exclusively based on special / personal characteristics, like their racial origin. Our venture was to apply anonymization techniques, such as Λ-anonymity, that they are part of Privacy Preserving techniques in order to confront the results of discrimination. The results of our practical implementation were pretty encouraging. We have concluded to the fact that the anonymization techniques help indeed to reduce the results of discrimination.