Awesome work! If I understand it correctly, it loads relevant subgraphs from the DB and then runs queries in Prolog on it? Or is it more similar to datalog?