Integrating induction and deduction for finding evidence of discrimination

Abstract

Automatic Decision Support Systems (DSS) are widely adopted for screening purposes in socially sensitive tasks, including access to credit, mortgage, insurance, labor market and other benefits. While less arbitrary decisions can potentially be guaranteed, automatic DSS can still be discriminating in the socially negative sense of resulting in unfair or unequal treatment of people. We present a reference model for finding (prima facie) evidence of discrimination in automatic DSS which is driven by a few key legal concepts. First, frequent classification rules are extracted from the set of decisions taken by the DSS over an input pool dataset. Key legal concepts are then used to drive the analysis of the set of classification rules, with the aim of discovering patterns of discrimination. We present an implementation, called LP2DD, of the overall reference model integrating induction, through data mining classification rule extraction, and deduction, through a computational logic implementation of the analytical tools.

Data preprocessing techniques for classification without discrimination

“…In Discrimination-Aware Data Mining two main directions can be distinguished: detection of discrimination [21][22][23][24], and the direction followed in this paper, namely learning classifiers if the data are discriminatory [4,15]. A central notion in the works on identifying discriminatory rules is that of the context of the discrimination.…”

A multidisciplinary survey on discrimination analysis

“…A GUI for visual exploratory analysis has been developed by Gao and Berendt (2011). Extensions of the approach by Pedreschi et al (2009);Ruggieri et al (2010a) deal with: (1) any measure built from a contingency table as in Figure 1; (2) the statistical significance of measures of discrimination of an extracted rule; (3) the discovery of affirmative actions and favoritism; and (4) the filtering of rules that can be supported by a genuine occupational requirement. The rankings imposed by the discrimination measures in Figure 1 are investigated by Pedreschi et al (2012): the choice of the reference measure critically affects the rankings of PD rules, with the RC (mainly adopted in the U.S.) and the RR (mainly adopted in the E.U.)…”

The Discovery of Discrimination

“…Confidence intervals and tests of statistical significance of the above measures are discussed in (Pedreschi et al, 2009;Ruggieri et al, 2010c). Here, we only mention that statistical tests will rank the rules according to how unlikely it is that they would be observed if there was equal treatment, not according to the severity of discrimination.…”

“…While a-protection is defined with reference to elift, its definition clearly applies to any measure from Figure 5.1. An extension of a-protection to account for its statistical significance is proposed in (Pedreschi et al, 2009;Ruggieri et al, 2010c). Also, we refer the reader to (Ruggieri et al 2010a(Ruggieri et al ,2010c for the presentation and experimentation of data mining algorithms able to efficiently extract a-protective classification rules from a large dataset of historical decision records.…”

“…An extension of a-protection to account for its statistical significance is proposed in (Pedreschi et al, 2009;Ruggieri et al, 2010c). Also, we refer the reader to (Ruggieri et al 2010a(Ruggieri et al ,2010c for the presentation and experimentation of data mining algorithms able to efficiently extract a-protective classification rules from a large dataset of historical decision records. Finally, (Pedreschi et al, 2012) show that the choice of a reference measure from Figure 5.1 has a critical impact on the ranking imposed over the set of PD classification rules.…”