Advancing AI Fairness: Addressing Gender Bias in Natural Language Processing Systems
Thesis submitted for the degree of Doctor of Philosophy, School of Computer Science, Technological University Dublin, December 2024.
Abstract
Natural Language Processing (NLP) systems have become integral to various real-world applications, but their effectiveness and fairness are heavily dependent on the quality and balance of their training data. A critical issue affecting these systems is gender bias, which can result in discriminatory outcomes and skewed performance across NLP tasks. This thesis explores the measurement and mitigation of gender bias in NLP systems, proposing novel approaches to ensure fairness and inclusivity. Four main contributions are made in this thesis.
This item has been moved.