Computing semantic relatedness is an essential operation for many natural language processing (NLP) tasks, such as Entity Linking (EL) and Question Answering (QA). It is still challenging to find a scalable approach to compute the semantic relatedness using Semantic Web data. Hence, we present for the first time an approach to pre-compute the semantic relatedness between the instances, relations, and classes of an ontology, such that they can be used in real-time applications.