This is an archive of an inactive wiki and cannot be modified.

Ways of using OWL KBs as data

This page attempts to clarify the different possibilities there as for accepting OWL Knowledge Bases as data of RIF rules.

The text on this page is intended to be a clarification of the Design Constraint The RIF Core must be able to accept OWL KBs as data. For a more detailed discussion on how OWL and (nonmonotonic) rules may interact, see (1).

There are two main ways of using OWL (DL) knowledge bases as data for RIF rules. These are: (1) using the OWL KB as external processor and (2) tight integration between OWL DL and rules, based on integration of models.

OWL KB as external query processor

The OWL KB is seen as a black box and "queries" are issued from the body of the rule (see (2)). The OWL KB is seen as an external query processor, see the design constraint RIF rules should be able to call out to external query processors.

This way of using OWL KBs combines straightforwardly with RIF-FOL, RIF-LP and RIF-PR.

There are questions as to (a) which kind of queries are allowed and (b) what kind of interaction from the rules to the OWL KB is allowed:

Kinds of queries

One can allow different kinds of queries to OWL (DL) knowledge bases. More specifically:

1. Class and property membership queries.

2. Negated class and property membership queries.

3. Subsumption queries. Checking whether a certain subsumption relation is entailed by the OWL KB.

4. Conjunctive query containment.

5. Arbitrary conjunctive queries.

Note that queries of the kinds 1-3 correspond to standard reasoning services of current Description Logic reasoning, whereas queries of the kinds 4-5 are not implemented for many Description Logics and there are no known effective algorithms for these tasks for expressive description logics, including OWL DL.

Interaction from rules to OWL KB

One may allow a flow of information from the rules to the OWL KB when evaluating the queries. One can could "add" facts, negated facts, or even more complex formulas.

Tight integration

A tighter integration of rules and OWL DL can be achieved by allowing the use of OWL DL class and properties directly in the rules and defining the semantics based on integrated models of the OWL knowledge base and the rule base (see (3)).

This way of using OWL KBs combines straightforwardly with RIF-FOL (as in SWRL) and with RIF-LP under the stable model semantics (as in (3)). It is not clear how this combination with RIF-PR or with RIF-LP under the well-founded semantics would work.

Such a tight integration is undecidable in general, but using certain restrictions on the use of variables in rules, decidability can be regained. There is a distiction between predicates defined in the DL knowledge base, called DL predicates, and predicates defined by the rules, called Datalog predicates. DL predicates are allowed to occur in the bodies and heads of the rules.

The "DL-safeness" restriction on variables requires every variable in a rule to occur in a positive Datalog atom in the body. Reasoning can in this case be reduced to standard reasoning services of reasoners for the stable model semantics and Description Logics.

The "weak DL-safeness" restriction requires all variables which occur in the head to occur in positive Datalog predicates in the body of the rule. Reasoning can in this case be reduced to standard reasoning services of reasoners for the stable model semantics, but non-standard reasoning techniques for Description Logics (containment of conjunctive queries in unions of conjunctive queries). The main advantage of the wea safeness restriction, compared with the safeness restriction, is that it allows to express conjunctive queries over the DL knowledge base in the body of a rule.

(1) Jos de Bruijn, Thomas Eiter, Axel Polleres, Hans Tompits: "On Representational Issues about Combinations of Classical Theories with Nonmonotonic Rules", DERI Technical Report 2006-05-29. To be published as an invited paper in the Proceedings of the First International Conference on Knowledge Science, Engineering and Management (KSEM'06), LNAI, Springer-Verlag.

(2) Thomas Eiter, Thomas Lukasiewicz, Roman Schindlauer, Hans Tompits: Combining Answer Set Programming with Description Logics for the Semantic Web. KR 2004: 141-151

(3) Riccardo Rosati DL+log: Tight Integration of Description Logics and Disjunctive Datalog In Proceedings of the Tenth International Conference on Principles of Knowledge Representation and Reasoning (KR 2006), 2006.