The increasing reliance on data for decision making has led to a number of techniques for automatic knowledge acquisition such as Formal Concept Analysis (FCA). FCA creates a lattice comprising partial order relationships between sets of object instances in a domain (extent) and their properties (intent). This is mapped onto a semantic knowledge structure comprising domain concepts with their instances and properties. However, this automatic extraction of structure from a large number of instances usually leads to a lattice which is too complex for practical use. Algorithms to reduce the lattice exist. However, these mainly rely on the lattice structure and are agnostic about any prior knowledge about the domain. In contrast, this paper uses existing domain knowledge encoded in a semantic ontology and a novel relevance index to inform the reduction process. We demonstrate the utility of the proposed approach, achieving a significant reduction of lattice nodes, even when the ontology only provides partial coverage of the domain of interest.