• Rich Knowledge Representation

1. Abstract

Rich Knowledge Representation use cases focus on the use of rules as (part of) a declarative knowledge representation formalism. In this setting, rules are often used in conjunction with other representation formalisms, such as ontologies, in order to extend their expressive power. A typical example is the use of rules to describe complex relationships between binary predicates which cannot be captured in ontology languages such as RDF and OWL. On the other hand, OWL is often used to capture information that typically cannot be expressed using rules (such as the existence of unnamed individuals), so compatibility between rules and RDF/OWL is of crucial importance.

2. Status

This is an abstraction of a number of use cases that focus on the use of rules for knowledge representation.

Edited by Ian Horrocks.

3. Links to Related Use Cases

• Automatically generated rules

• Decision making in Health Care

• Filling the holes of OWL

• Fuzzy Reasoning with Brain Anatomical Structures

• Information Integration with Rules and Taxonomies

• Labeling Brain Anatomical Structures in Digital Images

• Managing incomplete information

• Ontology Mapping with OWL and Rules

• Representing some levels of fuzzy rules with the help of datatype built-ins

• SW rules for Health Care and Life Sciences

• Frame-based representation, Inheritance of defaults, Reification

• Supporting the Reuse of Rules

4. Common themes

• The use of rules to capture (part of) a declarative model of the domain of discourse that can be updated and queried.
• The combination of rules with ontology languages in order to provide expressive power that goes beyond that which is typically provided by either formalism.
• Expressive knowledge representation techniques, such as fuzziness, incomplete information, semistructured data, reification, frames.

5. Benefits of Interchange

• As with ontologies, the value of such models is greatest when they explicate a common understanding of the domain that can be shared and reused.

6. Requirements on the RIF

• Compatibility with RDF and OWL. Rules are often used in conjunction with RDF and OWL ontologies to capture knowledge that cannot be expressed using either ontologies or rules alone.
• Declarative semantics. Rules (plus ontologies) are used to develop a model of the domain of discourse that can be updated and queried. Users need to be able to understand the meaning of such a model, and of queries against it, independently of any particular inference/query answering system.

7. Narrative

A typical (simplified) example is as follows:

A medical application uses an ontology to provide a model of human anatomy that can be used to describe procedures and, e.g., to categorise them w.r.t. their cost. Unfortunately, knowledge about interactions between binary predicates cannot be expressed in the ontology language being used (e.g., OWL), and this would prevent the correct categorisation of some procedures. For example, a fixation procedure applied to a part of the femur (e.g., the head of the femur) should be categorised (for costing purposes) as a fixation procedure applied to the femur.

In general, what is required in this example is to be able to express complex relationships between binary predicates. Such relationships can easily be captured using rules. E.g., a rule can be used to state that performed-on(x,y) and part-of(y,z) implies performed-on(x,z).

8. Commentary

Rules are well established as a declarative knowledge representation formalism. There are several different semantics for such rules, including LP-based semantics, but when rules are used in conjunction with ontologies a First Order semantics compatible with RDF and OWL is often assumed (see, e.g., SWRL).