D16.1v0.2 The WSML Family of Representation Languages

WSML Working Draft 25 October 2004

This version: http://www.wsmo.org/2004/d16/d16.1/v0.2/20041025/
Latest version: http://www.wsmo.org/2004/d16/d16.1/v0.2/
Previous version: http://www.wsmo.org/2004/d16/v0.2/20040926/

Editor:: Jos de Bruijn
Authors:: Jos de Bruijn; Holger Lausen; Dieter Fensel
Former Authors:: Douglas Foxvog; Eyal Oren
Reviewer:: Ian Horrocks

For printing and off-line reading, this document is also available in non-normative PDF version. Note that the documentation for the WSML XML syntax is not included in this PDF. Instead, it is included in the following three PDF documents: XMLSchemaWSML, XMLSchemaID, and XMLSchemaExpr.

Changelog

Compared with the previous version of this document (2004-09-26), the following changes have been made:

The major editorial changes are:

The deliverable was structured more clearly; the deliverable has been structured in a number of parts and several improvements have been made in the style sheet in order to make the printed version more readable.
All elements common between all WSML variants have been moved to the new Chapter 2 Common WSML Basics.
Appendix E. A list of all WSML keywords has been added; this appendix lists all WSML keywords, along with the WSML-variants in which they are included and references to the sections where they have been described.
The relationship with previous WSML d16.x deliverables has been moved to Appendix F.
The superscript used for describing the OWL- variants was eliminated (e.g. OWL Lite^- was replaced with OWL Lite-). The superscripts caused unwanted linespacings, even in standards-compliant browsers.
Section "Namespaces" (before the changelog) was removed
The XML Schema documentation is now also available as PDF and links to this documentation are located with the link to the PDF version of this document.
The section on relating WSML-Core to the WSMO use cases was removed. If this section is still required it should be added to D3.2, as an analysis of the application of the different WSML variants in the use cases.
Implementation efforts around WSML are moved to Chapter 11. Related Efforts.

The changes in the content are:

Identifiers for WSML elements are now optional. The semantics of omitted identifiers is described in Section 2.1.2 Identifiers in WSML.
The mappings between WSML-Core and OWL have been moved to the Chapter 9. Mapping to OWL.
The general namespaces for the variants have been changed; the fragment identifiers were removed.
The square brackets used for parameterized relations have been replaced with round brackets.
Attributes and relations are no longer disjoint. Attributes are a special kind of relation and thus an attribute definition can be further refined by a relation definition of the same identifier.
Attributes of different concepts are no longer disjoint. In the previous version of this deliverable, attributes were renamed before reasoning by concatenating it with the concept name to which it belongs. This is no longer the case.
Specific property of binary relations (e.g. symmetry, transitive) can now only be expressed in attribute definitions and no longer in relation definitions. Rationale for this is that only WSML-Core and WSML-DL (in contrast with all other WSML variants) restrict relations to being binary and most binary relations will anyway be written down as attributes.
The chapter 8 RDF Syntax for WSML now contains general considerations on encoding ontologies in RDF. Furthermore, Section 8.2 Representing WSML using RDF contains a number of possible directions for encoding WSML in RDF. These possible directions should be discussed.
There is one single namespace for referencing WSML elements. Each keyword defined in this document is implicitly defined in the namespace: http://www.wsmo.org/2004/wsml
Section 2.2.1: WSML Variant contains now the identifiers for the specific WSML variants, which deviate from the form the variant-specific namespaces used to have. The version number is no longer reflected and the deliverable number is also removed from teh URI.
Types of attributes are now optional. Will this pose problems for the BNF?
Features newly (compared with D2v1.0) introduced in WSML-Core and WSML-Flight are reflected in the XML syntax.
The XML syntax now contains a new top-level element <wsml> which can be used in combination with the 'variant' attribute to identify the WSML variant used for the specification.
Section 2.1.2 now (informally) defines the semantics of anonymous IDs
Descriptions of the syntax for goals, web services and mediators have been added.
All descriptions of the ontology elements in WSML-Core have been updated.
WSML-Flight has now been specified

Major issues not yet addressed (because of time constraints of the authors) in this version of the deliverable are:

Update of the mapping to OWL
Characterization of the allowed logical expressions in WSML-Core (closely related with the mapping of logical expressions to OWL)
Mapping human-readable syntax to XML
The logical expression syntax needs to be defined in a way similar to the definition of logical expressions in D2.
The XML example need to be updated to the revised XML syntax.
Appendix E should have a table listing all the keywords in the logical expressions

Major future work:

Update and inclusion of the WSML grammar which was originally developed for D2v1.0
Update of the WSML example in the appendix as soon as the syntax is fixed and making sure that all examples in the document use this parts of this integrated example.
Specification of the RDF syntax (once agreement is reached about the overall approach; see also discussion issues in the next section)
Specification of WSML-DL when the language is required
Specification of WSML-Rule once the required features of the language have become clear
Specification of WSML-Full once the approach for defining the semantics is clear
Specification of the semantics of all WSML variants

Discussion issues

The following are the discussion points raised in the previous version of this document (Sep 26th), along with the outcome of the discussions. The outcome of the discussions is reflected in the current version of the document.

The use of the ofDataType keyword. Should we just replace it with ofType?
Outcome of the discussion: ofDataType is replaced with ofType
The use of angle brackets for sets of values; why not use a comma-separated list?
Outcome of the discussion: brackets are used for all sets of values. Currently, curly brackets are proposed, but we might want to have round brackets '(' and ')'. Notice that in the new version, round brackets are used for cardinality constraints instead of curly brackets.
A special keyword for OWL-style universal value restrictions? Currently, abstract attributes cannot be specified in WSML-Core, because the semantics of ofType differs from the semantics of the OWL universal value restriction. The question is: do we need to have a special modeling element for this kind of restrictions or do we want to rely on logical expressions?
Outcome of the discussion: The new keyword impliesType is introduced. The semantics is equivalent to universal value restrictions in OWL.
A special keyword for complete class definitions: do we want it? A complete class definition can not have any attributes.
Outcome of the discussion: There was no clear outcome; in fact, there were not many opinions of on the issue. The only voiced opinion was Rubén who thought there should be an additional keyword. I (Jos) have proposed in this version not to introduce an additional keyword, but instead to explicitly explain the concept of a defined class and to say explicitly how to write down a defined class using a logical expression in Section 3.2.1 Concepts
WSML-Core vs. Web Service specifications
Outcome of the discussion: Web service specifications are a very important part of WSML and thus should be reflected in this document. In order to "sell" the ontology part of WSML as an ontology language, the ontology specification part will be repeated in D18, which has been moved to OMWG. [except that it is not in the OMWG activity sheet?]
The use of the wsmlVariant keyword for the specification of the variant and the version of the WSML variant being used for the specification.
Outcome of the discussion: wsmlVariant (Section 2.2.1) is a mandatory keyword to be used at the top of every WSML specification. Furthermore, each specification which uses the WSML XML syntax has the mandatory root element <wsml>, which has a mandatory attribute 'variant'. A proposal was to make the wsmlVariant keyword optional and let it default to 'wsml-core' when omitted. this should be further discussed
For WSML-Flight: More expressive logical expressions. Namely, we allow full Datalog with integrity constraints, default negation and the use of the equality and inequality symbols in the body of a rule. We no longer restrict the kind of logical expressions that can be written down. Therefore, WSML-Flight already contains a powerful rule language
Outcome of the discussion: WSML-Flight has no limitations on the form of the logical expressions. Furthermore, WSML-Flight allows n-ary relations.
The use of hasValue vs hasValues and ofType vs. ofTypeSet needs to be discussed, together with general cardinality constraints (see the WSML Flight Section 4.1.1 for a proposal for the specification of cardinality constraints.
Outcome of the discussion: There is no longer a distinction between single-valued and set-valued attributes. Thus the keywords hasValues and ofTypeSet have been removed. Instead, it is possible to explicitly state the cardinality of a property, see also Section 4.1.1.

The following discussion issues have arisen for this version of the document:

Making wsmlVariant optional and default to wsml-full? If this is the case, the client can just work with that part of the specification he can work with.
Section 8.2 Representing WSML using RDF contains a number of possible directions for encoding WSML in RDF. These possible directions should be discussed. Notice that I (Jos) have suggested to not layer on top of RDFS (see the section for my arguments).
Basic namespace for WSML: http://www.wsmo.org/2004/wsml
Namespaces for specific versions: http://www.wsmo.org/2004/wsml/vx.y
Identifiers for WSML variants: http://www.wsmo.org/2004/wsml/wsml-variant
One XML schema for all variants? Originally, the approach for the specification of the XML syntax for WSML was to create different schemas for each of the variants. However, by using the wsmlVariant keyword there is already a way to specify the desired WSML variant. Thus, it should be sufficient to have just one schema for all variants.
One BNF grammar for all variants? Similar to the previous discussion issue.
The examples for the postcondition in the goal specification (Section 2.4.1) was taken from d3.3. However, the meaning of the logical expression is not really clear. Should this formula really be in WSML-Full and is it correct that none of the variables is quantified?

Abstract

We introduce WSML, a family of formal representation languages with its roots in Description Logics, First-Order Logic and Logic Programming. The conceptual modeling elements of WSML are based on the meta-model of WSMO.

The WSML variants have increasing expressiveness, starting with the intersection of Description Logic and Horn Logic and ending with full First-Order Logic with non-monotonic extensions.

WSML-Core semantically corresponds with the intersection of Description Logic and Horn Logic, extended with extensive datatype support in order to be useful in practical applications. WSML-Core is fully compliant with a subset of OWL, albeit that the datatype support in WSML-Core is already beyond OWL, because the datatype support in OWL is very limited.
WSML-Flight extends WSML-Core with more intuitive value restrictions and cardinality constraints. WSML-Flight is the preferred ontology modeling language, because of its rich set of modeling primitives for modeling different aspects of attributes, such as value constraints and integrity constraints, and its rich logical language which allows for writing down arbitrary rules.
WSML-Rule extends WSML-Flight to a fully-fledged Logic Programming language, including function symbols and higher-order features of and HiLog and possible Transaction Logic.
WSML-DL extends WSML-Core to a fully-fledged Description Logic.
WSML-Full unifies all WSML variants under a common First-Order umbrella with non-monotonic extensions.

All WSML variants are described in terms of a normative human-readable syntax. Besides the human-readable syntax we provide an XML and an RDF syntax for exchange between machines. Furthermore, we provide a mapping to and from the OWL syntax for basic inter-operation with OWL ontologies through a common semantic subset of OWL and WSML.

PART I: PRELUDE

1. Introduction