The study of fairness-related aspects in data analysis is an active field of research, which can be leveraged to understand and control specific types of bias in decision-making systems. A major problem in this context is fair clustering, i.e., grouping data objects that are similar according to a common feature space, while avoiding biasing the clusters against or towards particular types of classes or sensitive features. In this work, we focus on a correlation-clustering method we recently introduced, and experimentally assess its performance in a fairness-aware context. We compare it to state-of-the-art fair-clustering approaches, both in terms of classic clustering quality measures and fairness-related aspects. Experimental evidence on public real datasets has shown that our method yields solutions of higher quality than the competing methods according to classic clustering-validation criteria, without neglecting fairness aspects.

When Correlation Clustering Meets Fairness Constraints

La Cava, L;Mandaglio, D;Tagarelli, A
2022-01-01

Abstract

The study of fairness-related aspects in data analysis is an active field of research, which can be leveraged to understand and control specific types of bias in decision-making systems. A major problem in this context is fair clustering, i.e., grouping data objects that are similar according to a common feature space, while avoiding biasing the clusters against or towards particular types of classes or sensitive features. In this work, we focus on a correlation-clustering method we recently introduced, and experimentally assess its performance in a fairness-aware context. We compare it to state-of-the-art fair-clustering approaches, both in terms of classic clustering quality measures and fairness-related aspects. Experimental evidence on public real datasets has shown that our method yields solutions of higher quality than the competing methods according to classic clustering-validation criteria, without neglecting fairness aspects.
2022
978-3-031-18839-8
978-3-031-18840-4
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.11770/349857
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? 0
social impact