(Added text on the informal usage of OWL 2 DL and OWL 2 Full)
|Line 187:||Line 187:|
== Acknowledgements ==
== Acknowledgements ==
Revision as of 17:16, 27 March 2009
- Document title:
- OWL 2 Web Ontology Language
Document Overview (Second Edition)
- W3C OWL Working Group (see Acknowledgements)
The OWL 2 Web Ontology Language, informally OWL 2, is an ontology language for the Semantic Web with formally defined meaning. OWL 2 ontologies provide classes, properties, individuals, and data values and are stored as Semantic Web documents. OWL 2 ontologies can be used along with information written in RDF, and OWL 2 ontologies themselves are primarily exchanged as RDF documents.
This document, part 1 of 13 in the OWL 2 document set, serves as an introduction to OWL 2 and the various other OWL 2 documents. It describes the various syntaxes for OWL 2, the different kinds of semantics, the defined profiles (sub-languages), and the differences between OWL 1 and OWL 2.
- Status of this Document
Ontologies are formalized vocabularies of terms, often covering a specific domain and shared by a community of users. They specify the definitions of terms by describing their relationships with other terms in the ontology. OWL 2 is an extension and revision of the OWL Web Ontology Language developed by the W3C Web Ontology Working Group and published in 2004. OWL 2 is being developed (and this document was written) by a follow-on group, the W3C OWL Working Group. OWL and OWL 2 are designed to facilitate ontology development and sharing via the Web, with the ultimate goal of making Web content more accessible to machines.
Figure 1 gives an overview of the OWL 2 language, showing its main building blocks and how they relate to each other. The ellipse in the center represents the abstract notion of an ontology, which can be thought of either as an abstract structure or as an RDF graph (see 2.1 Ontologies). At the top are various concrete syntaxes (see 2.2 Syntaxes) that can be used to serialize and exchange ontologies. At the bottom are the two semantic specifications that define the meaning of OWL 2 ontologies (see 2.3 Semantics).
Note that many users of OWL 2 will need only one syntax and one semantics; for them, this diagram could be much simpler, with only their one syntax at the top, their one semantics at the bottom, and rarely a need to see what's inside the ellipse in the center.
The conceptual structure of OWL 2 ontologies is defined in the OWL 2 Structural Specification document [OWL 2 Structural Specification]. This document uses UML [UML] to define the structural elements available in OWL 2, explaining their roles and functionalities in abstract terms and without reference to any particular syntax. It also defines the functional-style syntax, which closely follows the structural specification and allows OWL 2 ontologies to be written in a compact form.
Any OWL 2 ontology can also be viewed as an RDF graph. The relationship between these two views is specified by the Mapping to RDF Graphs document [OWL 2 RDF Mapping], which defines a mapping from the structural form to the RDF graph form, and vice versa. The OWL 2 Quick Reference Guide [OWL 2 Quick Guide] provides a simple overview of these two views of OWL 2, laid out side by side.
In practice, a concrete syntax is needed in order to store OWL 2 ontologies and to exchange them among tools and applications. The primary exchange syntax for OWL 2 is RDF/XML [RDF Syntax]; this is indeed the only syntax that must be supported by all OWL 2 tools (see Section 2.1 of the OWL 2 Conformance document [OWL 2 Conformance]).
While RDF/XML provides for interoperability among OWL 2 tools, other concrete syntaxes may also be used. These include alternative RDF serializations, such as Turtle [Turtle]; an XML serialization [OWL 2 XML]; and a more "readable" syntax used in several ontology editing tools [OWL 2 Manchester Syntax]. Finally, the functional-style syntax can also be used for serialization, although its main purpose is specifying the structure of the language [OWL 2 Structural Specification]. It must be emphasized, however, that OWL 2 tools are not required to support any of these alternative syntaxes.
|Name of Syntax||Specification||Status||Purpose|
|RDF/XML||Mapping to RDF Graphs,|
|Mandatory||Interchange (can be written and read by all conformant OWL 2 software)|
|OWL/XML||XML Serialization||Optional||Easier to process using XML tools|
|Functional Syntax||Structural Specification||Optional||Easier to see the formal structure of ontologies|
|Manchester Syntax||Manchester Syntax||Optional||Easier to read/write DL Ontologies|
|Turtle||Mapping to RDF Graphs,|
|Optional, Not from OWL-WG||Easier to read/write RDF triples|
The OWL 2 Structural Specification document defines the abstract structure of OWL 2 ontologies, but it does not define their meaning. The Direct Semantics [OWL 2 Direct Semantics] and the RDF-Based Semantics [OWL 2 RDF-Based Semantics] provide two alternative ways of assigning meaning to OWL 2 ontologies, with a correspondence theorem providing a link between the two. These two semantics are used by reasoners and other tools to answer queries about, e.g., class consistency, subsumption and instance retrieval.
The Direct Semantics assigns meaning directly to ontology structures, resulting in a semantics compatible with the model theoretic semantics of the SROIQ description logic—a fragment of first order logic with useful computational properties. The advantage of this close connection is that the extensive description logic literature and implementation experience can be directly exploited by OWL 2 tools. However, some conditions must be placed on ontology structures in order to ensure that they can be translated into a SROIQ knowledge base, for example, transitive properties cannot be used in number restrictions (see Section 3 of the OWL 2 Structural Specification document [OWL 2 Structural Specification] for a complete list of these conditions). Ontologies that satisfy these syntactic conditions are called OWL 2 DL ontologies. Note that, informally, "OWL 2 DL" often refers to the syntactic subset and interpreted using the Direct Semantics [OWL 2 Direct Semantics]
The RDF-Based Semantics assigns meaning directly to RDF graphs and so indirectly to ontology structures via the Mapping to RDF-graphs. The RDF-Based Semantics is fully compatible with RDF Semantics [OWL 2 RDF-Based Semantics], and extends the semantic conditions defined for RDF. The RDF-Based Semantics can be applied to any OWL 2 Ontology, without restrictions, as any OWL 2 Ontology can be mapped to RDF. Informally, "OWL 2 Full" is often used to refer to RDF graphs considered as OWL 2 ontologies and interpreted using the RDF-Based Semantics [OWL 2 RDF-Based Semantics].
The correspondence theorem in Section 7.3 of the RDF-Based Semantics Document [OWL 2 RDF-Based Semantics]) defines a precise, close relationship between the Direct and RDF-Based Semantics. This theorem states, in essence, that given an OWL 2 DL ontology, inferences drawn using the Direct Semantics will still be valid if the ontology is mapped into an RDF graph and interpreted using the RDF-Based Semantics.
OWL 2 Profiles [OWL 2 Profiles] are sub-languages (syntactic subsets) of OWL 2 that offer important advantages in particular application scenarios. Three different profiles are defined: OWL 2 EL, OWL 2 QL, and OWL 2 RL. Each profile is defined as a syntactic restriction of the OWL 2 Structural Specification, i.e, as a subset of the structural elements that can be used in a conforming ontology, and each is more restrictive than OWL DL. Each of the profiles trades off different aspects of OWL's expressive power in return for different computational and/or implementational benefits.
OWL 2 EL enables polynomial time algorithms for all the standard reasoning tasks; it is particularly suitable for applications where very large ontologies are needed, and where expressive power can be traded for performance guarantees. OWL 2 QL enables conjunctive queries to be answered using standard relational database technology; it is particularly suitable for applications where relatively lightweight ontologies are used to organize large numbers of individuals and where it is useful or necessary to access the data directly via relational queries (e.g., SQL). OWL 2 RL enables the implementation of polynomial time reasoning algorithms using rule-extended database technologies operating directly on RDF triples; it is particularly suitable for applications where relatively lightweight ontologies are used to organize large numbers of individuals and where it is useful or necessary to operate directly on data in the form of RDF triples.
Any OWL 2 EL, QL or RL ontology is, of course, also an OWL 2 ontology and can be interpreted using either the Direct or RDF-Based Semantics. When using OWL 2 RL, a rule-based implementation can operate directly on RDF triples and so can be applied to an arbitrary RDF graph, i.e., to any OWL 2 ontology. In this case, reasoning will always be sound (that is, only correct answers to queries will be computed), but it may not be complete (that is, it is not guaranteed that all correct answers to queries will be computed). Theorem PR1 of the Profiles document states, however, that (in general) when the ontology is consistent with the structural definition of OWL 2 RL, a suitable rule-based implementation performing ground atomic queries will be both sound and complete.
3 Differences between OWL 1 and 2
The 2004 version of OWL [OWL 1 Overview] (referred to hereafter as “OWL 1”) has a very similar overall structure to OWL 2. Looking at Figure 1, almost all the building blocks of OWL 2 were present in OWL 1, albeit under a possibly different name.
- The role of the structure, as well as the functional syntax of OWL 2, was played by the OWL Abstract Syntax [Section 2 of OWL 1 Semantics]. The two syntaxes are different but their respective role within the overall structure of OWL is identical. The new functional syntax is much closer to the RDF graph representation and can capture more RDF graphs; it also has a formal equivalence to UML [UML].
- Like OWL 1, OWL 2 specifies a precise mapping from ontology structures (represented using the abstract/functional syntax) to RDF graphs. OWL 2, however, also benefits from an explicitly specified mapping from RDF graphs back to ontology structures.
- The two semantics (RDF-Based and Direct) of OWL 2 have their direct counterparts in OWL 1, under the name of RDF-Compatible Model-Theoretic Semantics [Section 5 of OWL 1 Semantics] and Direct Model-Theoretic Semantics [Section 3 of OWL 1 Semantics].
- An XML Presentation Syntax was also available for OWL 1 [OWL 1 XML Syntax] (although not as a Recommendation). On the other hand, the Manchester syntax did not exist for OWL 1.
- OWL 1 defined one sub-lanuage (OWL Lite), where OWL 2 defines three (EL, QL, and RL). OWL Lite has not been re-specified for OWL 2, but because of backward compatibility, OWL Lite ends up as a sub-language of OWL 2.
The central role of RDF/XML, the role of other syntaxes, and the relationships between the Direct and RDF-Based semantics (i.e., the correspondence theorem) have not changed. More importantly, the changes are backward compatible: all OWL 1 Ontologies remain valid OWL 2 Ontologies, with identical inferences (except in certain minor cases, detailed in 3.2 Modifications).
3.1 New Features
OWL 2 adds new functionality with respect to OWL 1. Some of the new features are syntactic sugar (e.g., disjoint union of classes) while others offer new expressivity, including:
- property chains;
- richer datatypes, data ranges;
- qualified cardinality restrictions;
- asymmetric, reflexive, and disjoint properties; and
- enhanced annotation capabilities
The OWL 2 New Features and Rationale document [OWL 2 New Features and Rationale] describes all the new features of OWL 2, while the OWL 2 Quick Reference Guide [OWL 2 Quick Guide] provides an overview of the various OWL 2 features in general, clearly indicating those that are new.
The restrictions applicable to OWL DL have also been relaxed somewhat in OWL 2. Whereas, in OWL 1, there was a very strict separation between, for example, classes and individuals, OWL 2’s corresponding restrictions are a bit more relaxed. Under some circumstances the same identifier (i.e., URI) can be used to denote, for example, both a class and an individual while still maintaining the usability of Direct Semantics. In other words, the set of RDF Graphs that can also be handled by Description Logics reasoners has become a bit larger in OWL 2 compared to OWL 1.
4 Documentation Roadmap
The OWL 2 ontology language is normatively defined by five core specification documents describing its conceptual structure, primary exchange syntax (RDF/XML), two alternative semantics (Direct and RDF-Based), and conformance requirements. Three additional specification documents describe optional features that may be supported by some implementations: the language profiles, and two alternative concrete syntaxes (XML and Manchester).
These documents are, however, all rather technical and mainly aimed at OWL 2 implementers and tool developers. Those looking for a more approachable guide to the features and usage of OWL 2 may prefer to consult one of the user documents (Primer, New Features and Rationale, and Quick Reference Guide).
Note: the Manchester Serialization Syntax and the Data Range Extension documents are to be published as W3C Working Group Notes. All other documents are published as W3C Recommendations.
The regular attendees at meetings of the OWL Working Group at the time of publication of this document were: Jie Bao (RPI), Diego Calvanese (Free University of Bozen-Bolzano), Bernardo Cuenca Grau (Oxford University Computing Laboratory), Martin Dzbor (Open University), Achille Fokoue (IBM Corporation), Christine Golbreich (Université de Versailles St-Quentin and LIRMM), Sandro Hawke (W3C/MIT), Ivan Herman (W3C/ERCIM), Rinke Hoekstra (University of Amsterdam), Ian Horrocks (Oxford University Computing Laboratory), Elisa Kendall (Sandpiper Software), Markus Krötzsch (FZI), Carsten Lutz (Universität Bremen), Deborah L. McGuinness (RPI), Boris Motik (Oxford University Computing Laboratory), Jeff Pan (University of Aberdeen), Bijan Parsia (University of Manchester), Peter F. Patel-Schneider (Bell Labs Research, Alcatel-Lucent), Sebastian Rudolph (FZI), Alan Ruttenberg (Science Commons), Uli Sattler (University of Manchester), Michael Schneider (FZI), Mike Smith (Clark & Parsia), Evan Wallace (NIST), Zhe Wu (Oracle Corporation), and Antoine Zimmermann (DERI Galway). We would also like to thank past members of the working group: Jeremy Carroll, Jim Hendler, and Vipul Kashyap.