This document presents an approach for exception-enriched rule learning from knowledge graphs. It aims to mine rules from knowledge graphs while accounting for exceptions to the rules. The approach involves: 1) Mining Horn rules (if-then rules) from a knowledge graph 2) Extracting an exception witness set (EWS) of entities that violate the mined rules 3) Constructing candidate rule revisions by adding exceptions to the rules 4) Selecting the best rule revision by maximizing prediction accuracy while minimizing conflicting predictions against the knowledge graph. The goal is to learn more accurate rules from knowledge graphs by incorporating known exceptions.