The potential of prescribing efficacious medications to improve patient outcomes and reduce side effects has garnered significant interest. Ontology describes the various kinds of entities that can be found in a domain, along with their characteristics and interrelationships. Extraction from Ontology through RDF Query Language (SPARQL) and SPARQL Protocol relies completely on the static data in the ontology. We have proposed a novel approach that leverages the combination of Ontology and deep learning domains. Our work involves building an ontology from electronic health records (EHR) and utilizing SPARQL to retrieve the medications that can be recommended. We are using deep learning to compute effectiveness score by doing sentimental analysis on drug reviews with a 96% accuracy rate in order to provide medications with fewer side effects. Finally the results of SPARQL will be compared with effectiveness scores obtained to recommend more effective and less adverse drug for a patient.