introduction to description logics and owl
DESCRIPTION
Introduction to Description Logics and OWL. Nick Gotts & Gary Polhill. What is an ontology?. “…a formal, explicit specification of a shared conceptualisation” [Gruber, 1993; Fensel, 2001] Formal: machine readable Explicit: types of concepts and constraints on their use - PowerPoint PPT PresentationTRANSCRIPT
Craigiebuckler, Aberdeen, AB15 8QH, UK
Introduction to Description Logics and OWL
Nick Gotts & Gary Polhill
What is an ontology?• “…a formal, explicit specification of a shared
conceptualisation”[Gruber, 1993; Fensel, 2001]
– Formal: machine readable– Explicit: types of concepts and constraints on their use– Shared: consensual knowledge accepted by a group– Conceptualisation: an abstract model
• Essentially contain the following:– Classes of concept with declarative conditions for class
membership– Relationships between classes
Gruber, T. R. (1993) Knowledge Acquisition 5, pp. 199-220.Fensel, D. (2001) Ontologies: A Silver Bullet for Knowledge Management and Electronic Commerce. Springer
What can they do for CAVES?
• CAVES: Not one model but many– One unifying ontology may not be possible or
desirable• Comparison of ontologies of different case
studies• Comparison of ontologies at different levels of
granularity• Capturing relationships between ontologies• Higher-level descriptions of models that are
nevertheless formal
Owl and Description Logics
• OWL (Web Ontology Language)– Allows ontologies to be published on the web– Allows ontologies to link to each other– One representation uses XML (via RDF)– Used in Protégé (ontology editor)– Conceptually linked to Description Logics
• Description Logics– Origins– Some key properties– Reasoning within Description Logics.– Description Logics and the various versions of OWL.
Owl
• Web Ontology Language
• Relation to XML and RDF [CHECK!]
• Origins and history
Logics• Formal languages with inference rules• Types of logic:
– Propositional logic– First-order predicate logic– Higher-order predicate logics– Modal logics– Temporal logics– …– Description logics
• Trade off: expressivity against computational tractability
Description Logics
• Descended from AI approaches to knowledge representation (semantic nets, frames)– But with a formal, logic-based semantics
• Concepts and roles• What sorts of things can be expressed in description
logics? [EXAMPLES]– Description formalism– Terminological formalism– Assertional formalism
Reasoning in Description Logics
• Properties of Logics– Completeness– Decideability
• Properties of algorithms for decideable logics:– Worst-case time complexity– Typical case / “In-practice” time complexity
• Specialised automated reasoners for Description Logics: tableau-based algorithms
Description Logics as Ontology Languages• Several web ontology languages, including OWL, use the
Description Logic SHIQ as basis of their design.• Ontologist-friendly features of SHIQ :
– Qualified number restrictions– Complex terminological axioms– Inverse roles, transitive roles, subroles
• Reasoning in SHIQ :– Decideable– Worst-case time-complexity exponential– Highly optimized SHIQ reasoners, e.g. RACER, behave quite well
in practice• ?Extensions to SHIQ :
– Concrete domains– Nominals
Description of SHIQ ?
– Tbox and Abox– Verifying the TBox (p.13)– Tableau-based decision procedure
(p.16) “nondeterminisitc double exponential time” [!?!]