Universidade Federal do Ceará
Mestrado e Doutorado em Ciências da Computação

Título: A comprehensive survey on two-sided fairness in recommendation systems

Data: 31/10/2025
Horário: 10h
Local: Sala de Seminários / Bloco 952

Resumo:
Recommendation systems have become essential tools for assisting users in discovering content that aligns with their preferences while helping organizations enhance engagement. Despite their benefits in
providing personalized experiences and supporting business growth, these systems may reinforce or amplify existing biases. Such biases can lead to unequal treatment across demographic groups or favor
certain users and item providers (e.g., sellers and content creators) over others. One of the major challenges in recommendation systems is the mitigation of biases that are introduced or propagated by algorithms, potentially harming both customers and providers. In this survey, we focus on two-sided fairness, a research direction that aims to ensure fair treatment for both stakeholders, users and item providers, simultaneously. We comprehensively review the literature that proposes solutions to balance fairness across both sides of the platform. Furthermore, we develop a taxonomy that categorizes approaches according to their adopted fairness definition employed for each stakeholder, considering their level of granularity (individual- or group-level). By analyzing these methods, we identify their objectives and their employed strategies. Additionally, we provide insights drawn from research questions and comparative experimental evaluations that reveal impacts of achieving fairness for both stakeholders. Finally, we discuss the open challenges and outline directions for future research toward building a fair recommendation ecosystem.

Banca examinadora:
Prof. Dr. Javam de Castro Machado (MDCC/UFC) – Orientador
Prof. Dr. Angelo Roncalli Alencar Brayner (MDCC/UFC)
Prof. Dr. César Lincoln Cavalcante Mattos (MDCC/UFC)
Prof. Dr. Eliezer de Souza da Silva (MDCC/UFC)