An approximation algorithm for querying inconsistent knowledge bases

•We discuss, in an educational and easy-to-read way, a technique for repairing inconsistent clinical knowledge bases by using a more fine-grained repair primitive which updates facts rather than deleting them.•We show the notion of a universal repair, which compactly represents the whole set of possible repairs and it is useful to address the problem of computing approximate answers in inconsistent clinical knowledge bases.•We formally present a sound polynomial-time approximation algorithm for computing consistent query answers making use of universal repairs.•We provide examples showing that our approximation schema yields strictly better results than wellknown approximation approaches.