How to best perform a search over a dataset is a common field of research in the scientific community. Query engines are used when the information is handled by an ontology in order to obtain structured information semantically. A common issue that arises is when the same query is performed several times, as the query engine must check the domain every time in order to retrieve the information. In this paper, we propose an architecture that takes advantage of the concept of Reflexive Ontologies (RO) in order to achieve timely semantic retrieval. The proposed architecture is illustrated by a case study in the Film Heritage domain, showing a performance improvement and providing ground for further research and discussion.