The document presents a novel approach for entity linking that combines a semantic connectivity score (SCS) and a co-occurrence-based measure (CBM). SCS measures relatedness of entity pairs in knowledge graphs using Katz index, while CBM approximates co-occurrence in web resources. Evaluation on news documents shows the combined approach improves precision and recall over individual methods. The authors conclude it provides a scalable entity linking technique that correctly links entities marked as unrelated by human evaluators.