Retrieval as a practical fix for data fragmentation in Africa

A benchmark study of Retrieval-Augmented Generation (RAG) for African institutional data, demonstrating how retrieval grounding mitigates contextual blindness in large language models and improves accuracy, faithfulness, and trustworthiness across fragmented, multilingual, and heterogeneous information sources.