k-NN as an Implementation of Situation Testing for Discrimination Discovery and Prevention

Luong, Binh Thanh and Ruggieri, Salvatore and Turini, Franco (2011) k-NN as an Implementation of Situation Testing for Discrimination Discovery and Prevention. Technical Report del Dipartimento di Informatica . Università di Pisa, Pisa, IT.

Other (GZip) Available under License Creative Commons Attribution No Derivatives. Download (377Kb)

Abstract

<p>With the support of the legally-grounded methodology of situation<br />testing, we tackle the problems of discrimination discovery and<br />prevention from a dataset of historical decisions by adopting a<br />variant of k-NN classification. A tuple is labeled as<br />discriminated if we can observe a significant difference of<br />treatment among its neighbors belonging to a protected-by-law<br />group and its neighbors not belonging to it. Discrimination<br />discovery boils down to extracting a classification model from the<br />labeled tuples. Discrimination prevention is tackled by changing<br />the decision value for tuples labeled as discriminated before<br />training a classifier. The approach of this paper overcomes legal<br />weaknesses and technical limitations of existing proposals.</p>

Repository staff only actions