We demonstrate an approach for enriching text-derived knowledge bases with biological detail by incorporating cell type context into protein-protein interaction networks at PubMed scale