Database Reference
In-Depth Information
Chapter 5
The Discovery of Discrimination
Dino Pedreschi, Salvatore Ruggieri, and Franco Turini
Abstract.
Discrimination discovery from data consists in the extraction of dis-
criminatory situations and practices hidden in a large amount of historical deci-
sion records. We discuss the challenging problems in discrimination discovery, and
present, in a unified form, a framework based on classification rules extraction and
filtering on the basis of legally-grounded interestingness measures. The framework
is implemented in the publicly available DCUBE tool. As a running example, we
use a public dataset on credit scoring.
5.1
Introduction
Human right laws (European Union Legislation, 2011; United Nations Legisla-
tion, 2011; U.S. Federal Legislation, 2011) prohibit discrimination against protected
groups on the grounds of race, color, religion, nationality, sex, marital status, age and
pregnancy; and in a number of settings, including credit and insurance; sale, rental,
and financing of housing; personnel selection and wages; access to public accom-
modations, education, nursing homes, adoptions, and health care. Several authorities
(regulation boards, consumer advisory councils, commissions) monitor and report
on discrimination compliances. For instance, the European Commission publishes
an annual report on the progress in implementing the Equal Treatment Directives
by the member states (see Chopin & Do, 2010); and in the US the Attorney General
reports to the Congress on the annual referrals to the Equal Credit Opportunity Act.
Given the current state of the art of decision support systems (DSS), socially
sensitive decisions may be taken by automatic systems, e.g., for screening or rank-
ing applicants to a job position, to a loan, to school admission and so on. Classi-
cal approaches adopted in legal cases (Finkelstein & Levin, 2001) are limited to
the verification of an hypothesis of possible discrimination by means of statistical
ยท
Search WWH ::
Custom Search