Big Data, Algorithmic Bias, and the Future of Library Recommendation Systems in Nigerian Academic Libraries

This study critically examined the implications of algorithmic bias in library recommendation systems for equitable information access in Nigerian academic libraries. As academic libraries begin adopting automated recommendation tools powered by machine learning, questions arise about whether these systems replicate or amplify social inequalities in information retrieval. The study employed a mixed critical-technical design: algorithmic audit methodologies were applied to three library recommendation systems deployed in Nigerian university libraries, combined with qualitative interviews with 35 library users from underrepresented groups across universities in Kano, Lagos, and Enugu. Findings demonstrated that recommendation outputs systematically over-represented English-language, Global North-authored publications and underrepresented African, Nigerian, and female-authored scholarship, reflecting biases embedded in training data. Indigenous language materials were almost entirely absent from recommendation outputs. The study developed an original framework for bias-aware library recommendation system design incorporating local knowledge equity indicators. It makes a theoretical contribution to critical library and information science scholarship and to AI ethics in the African context. Keywords: algorithmic bias, library recommendation systems, big data, Nigerian academic libraries, information equity.