[tannhaeuser]

RDF, and in particular OWL2 (reformulation of RDF tech based on description logic) is about decidable fragments of first-order logic, whereas Prolog is an existential Horn fragment on terms with Turing-complete extra-logical additions such as negation-as-failure and "cut" hence has undecidable decision problems. I actually think the EU-granted research belittled in another comment did a good job of carving out fragments of FOL for relevant applications with desirable complexities of decision problems. But from a practical PoV, RDF, OWL, and SPARQL is bordering on unusable (starting from the fact that open world semantics alone isn't applicable in many real-world scenarios), though jena and rdflib are working fine. Think about OWL2/description logic as a variable-free representation of axioms with just two logical variables (plus some with three vars such as the axiom of transitivity).

Today there's a renewed interest in Prolog and Datalog, which makes me happy after RDF had captured the field for almost two decades.