4 Replies Latest reply on Mar 25, 2008 10:00 PM by Mannamal-Oracle

    Referential Integrity

    18661
      Hi,
      Is there a way to ensure semantic consistency of the ontology stored in Oracle 11g at insert/update/delete time? This would be equivelant to referential integrity in the relational model.

      Thanks,
      Essam
        • 1. Re: Referential Integrity
          Mannamal-Oracle
          Hi Essam,

          Validation is possible, see section 2.2.5 in the documentation for details, but the validate procedure is something that can be executed after the insert/update/delete operations. That is, the validation procedure is something the user or the application specfically calls and is not done automatically.

          If your question addressed something else, could you please elaborate on what you mean by semantic consistency?

          Melli
          • 2. Re: Referential Integrity
            18661
            Hi Elli,
            Thank you for your reply. I was hoping the validation would be done at the time the insert/update/delete operation is executed, similar what the database referential integrity constraints is being checked. I think this would keep the data "semantically consistent" with the restrictions defined in the ontology. Consequently, such real-time validation would make the RDF/OWL oracle store suitable for transactional usage as well.

            Thanks,
            Essam
            • 3. Re: Referential Integrity
              24743
              It should be possible to automatically validate triplets against an ontology (RDFS/OWL) f.ex. super/sub classes, domains, ranges, cardinalities, data types, ...

              Are there plans for implementing this?

              Regards - Christian
              • 4. Re: Referential Integrity
                Mannamal-Oracle
                Reponse to Essam: One possibility is to call the validation procedure via a trigger on inserts. But of course that would be computationally intensive. But perhaps there are only some situations that could lead to semantic inconsistency, and the trigger could be used for only those situations (and not on every insert)?

                Response to Christian: This is something we can look at in the future. If you have more detailed feedback on what you like to see do write to me at melliyal <dot> annamalai <at> oracle <dot> com

                Melli