I’ve written a post exploring the fundamentals of Information Retrieval and how they relate to modern RAG (Retrieval-Augmented Generation) systems. It walks through: • The CISI dataset used for experiments • Sparse retrieval methods — TF-IDF and BM25, with their underlying mechanics • Evaluation metrics — MRR, Precision@k, Recall@k, and NDCG • Vector-based retrieval with embedding models • ColBERT and late-interaction (MaxSim) methods