%@page %> <%-- This file is the text of the working draft. It excludes the stuff before the TOC; the TOC; and the tests themselves. TODO for making editors PR - delete proposed.zip - delete proposed appendix --%> <%=head(1,"introduction","Introduction")%>
As part of the definition of the Web Ontology Language (OWL) the Web Ontology Working Group provides a set of test cases. This document presents those test cases. They are intended to provide examples for, and clarification of, the normative definition of OWL found in <%=ref("OWL Semantics and Abstract Syntax")%> to which this document is subsidiary.
This document is one component of the description of OWL, the Web Ontology Language, being produced by the W3C Web Ontology Working Group. The Document Roadmap section of the <%=ref("OWL Overview")%> describes each of the different parts and how they fit together.
This document describes the various types of test used
and the format in which the tests
are presented.
Alternative formats of the test collection are provided.
These are intended to be suitable
for use by OWL developers in test harnesses,
possibly as part of a test driven development process,
such as Extreme Programming <%=ref("XP")%>.
The format of the Manifest
files
used as part of these alternative formats is described.
This document describes the process for conflict resolution and errata related to these tests.
--%>In the non-normative appendices, this document also describes the process for creation and approval of these tests.
<%--Further appendices show further proposed tests.
--%> <%=head(2,"scope","Conformance and Scope")%>Various conformance levels are defined in this document in terms of <%=ref("OWL Semantics and Abstract Syntax")%>.
However, the test cases do not constitute a conformance test suite for OWL, since they are silent on several important issues. This document cannot be considered a complete specification of OWL.
The tests illustrate issue resolutions, and illustrate the use and meaning of the terms in the OWL namespace.
There are other miscellaneous tests: some arising in the literature, and in preexisting systems; others intending to show the difficulty of complete implementations of OWL Full.
<%=head(1,"deliverables","Deliverables (Normative)")%>The deliverables included as part of the test cases are:
Note: Other files can be found under the top URL of the Web site which are not part of the deliverable.
<%=head(2,"normativity","Normative Status")%>Of the deliverables the only normative tests are those included in this document. All other deliverables are informative. Moreover, the recommendation document is informative except for the conformance statements, the test data (specified in RDF/XML <%=ref("RDF Syntax")%>), and the supporting documentation.
<%--The Web Ontology Working Group has seen adequate implementation experience of most of the tests in this document. Some, however, are particularly difficult to implement efficiently. These are labelled as extra credit tests. Such tests indicate the semantics of OWL, but may use features that are not sufficiently widely implemented to provide good interoperability.
A general case of extra credit tests is that all OWL Full nonentailments and consistency tests are extra credit tests. This is because typical OWL Full implementations prove entailments but cannot prove nonentailments.
Extra credit tests are labelled with "EC" within this document and with status EXTRACREDIT in the manifest files.
The name indicates that there is no expectation that any implementation will successfully run such tests and any that do gain extra credit.
<%--
Each test consists of one or more RDF/XML documents and a Manifest
file.
Tests of one document indicate some property of that document
when viewed as an OWL knowledge base.
Tests of two or more documents indicate a relationship between the two documents
when viewed as OWL knowledge bases.
The Manifest
file is named <%=named("Manifest")%> (The NNN
is replaced by the test number).
It contains metadata (in RDF) indicating the test type,
and describing the test.
The metadata also indicates the language levels appropriate for each test and each document in each test. For each RDF/XML document, one language level is indicated, being OWL Lite, OWL DL or OWL Full, as given by the syntactic rules in <%=ref("OWL Semantics and Abstract Syntax")%>. For semantic tests, one or two language levels are indicated. If the language level OWL Full is indicated for a semantic test, then the test holds according to the RDF-Compatible Model-Theoretic Semantics in <%=ref("OWL Semantics and Abstract Syntax")%>. If the language level OWL Lite or OWL DL is indicated for a semantic test, then the test holds according to the Direct Model-Theoretic Semantics in <%=ref("OWL Semantics and Abstract Syntax")%>. If the language level OWL Lite is indicated for a semantic test, then the test only uses features within the OWL Lite sublanguage.
Some of the tests require that certain
datatypes are, or are not, supported in the
datatype map
<%=ref("OWL Semantics and Abstract Syntax")%>.
These are indicated with the test.
Other datatypes which are used in the test
are also indicated: the test applies whether or not these are supported in the
datatype map .
The datatypes
xsd:integer
, xsd:string
from <%=ref("XML Schema Datatypes")%>
are not indicated, even when used or required, since they
must be supported.
These tests use one document. It is named <%=named("bad")%>. This document includes a use of the OWL namespace with a local name that is not defined by the OWL recommendation. An OWL Syntax checker SHOULD give a warning.
Note: These tests are intended to help migration from DAML+OIL <%=ref("DAML+OIL")%>, since the local names chosen are defined in the DAML+OIL namespace.
<%=head(2,"testEntailment","Entailment Tests")%>
These tests use two documents.
One is named <%=named("premises")%>,
the other is named <%=named("conclusions")%>.
The conclusions
are
entailed by the premises
.
Such entailment is defined by the OWL semantics <%=ref("OWL Semantics and Abstract Syntax")%>,
(see also
OWL Full entailment).
These tests use two documents.
One is named <%=named("premises")%>,
the other is named <%=named("nonconclusions")%>.
The nonconclusions
are not
entailed
by the premises
.
Such entailment is defined by the OWL semantics <%=ref("OWL Semantics and Abstract Syntax")%>,
(see also
OWL Full entailment).
Exceptionally, test imports-002 includes a third document.
<%=head(2,"testTrue","True Tests")%>
These tests use one document.
It is named <%=named("conclusions")%>.
The conclusions
follow from the OWL semantics
<%=ref("OWL Semantics and Abstract Syntax")%>.
These tests are a special case of the entailment tests
in which the premises are empty.
These tests use one document.
It is named <%=named("conclusions")%>.
These are a special case of true tests.
The conclusions
follow from the
OWL Full semantics
<%=ref("OWL Semantics and Abstract Syntax")%>.
The tests are intended to illustrate how
OWL Full can be used to describe its own properties and
classes.
These tests use one document. It is named <%=named("consistent")%>. The document is consistent as defined by the OWL Semantics <%=ref("OWL Semantics and Abstract Syntax")%>, (see also OWL Full consistency).
<%=head(2,"testInconsistency","Inconsistency Tests")%>These tests use one document. It is named <%=named("inconsistent")%>. The document is not consistent as defined by the OWL semantics <%=ref("OWL Semantics and Abstract Syntax")%>, (see also OWL Full consistency).
<%=head(2,"testImportEntailment","Import Entailment Tests")%>
These tests use more than two documents.
One is named <%=named("premises")%>,
another is named <%=named("conclusions")%>, the rest have names
like supportNNN-A.rdf
.
The support
documents are in the
imports closure of the
premises
document.
The conclusions
are
entailed
by the
imports closure
of the premises
.
Such entailment is defined by the OWL semantics <%=ref("OWL Semantics and Abstract Syntax")%>,
(see also
OWL Full entailment).
These tests use two documents.
One is named <%=named("imports")%>,
the other is named <%=named("main")%>.
These
tests indicate the
interaction between owl:imports
and the sublanguage levels of the main
document.
An OWL Full document is any RDF/XML document <%=ref("RDF Syntax")%>.
An OWL DL document is an OWL Full document such that the imports closure <%=ref("OWL Semantics and Abstract Syntax")%> of the corresponding RDF graph <%=ref("RDF Concepts")%> is an OWL DL ontology in RDF graph form.
An OWL Lite document is an OWL Full document such that the imports closure <%=ref("OWL Semantics and Abstract Syntax")%> of the corresponding RDF graph <%=ref("RDF Concepts")%> is an OWL Lite ontology in RDF graph form.
<%=head(3,"consistencyConformance","Semantic Conformance")%> <%-- OLD TEXTAn OWL document is consistent with respect to a datatype theory <%=ref("OWL Semantics and Abstract Syntax")%>, if and only if there exists some model of the document that is consistent with the constraints specified by the relevant model theory (see <%=ref("OWL Semantics and Abstract Syntax")%>: OWL Lite and OWL DL, OWL Full).
--%>An OWL Lite or OWL DL document D is OWL DL consistent with respect to a datatype map T if and only if there is some abstract OWL interpretation I with respect to T such that I satisfies an abstract ontology O corresponding to D, in which O has a separated vocabulary; (see <%=ref("OWL Semantics and Abstract Syntax")%>).
An OWL Full document D is OWL Full consistent with respect to a datatype map T, if and only if there is some OWL Full interpretation I with respect to T such that I satisfies all the RDF graphs in some imports closed collection containing an RDF graph corresponding to D.
<%=head(2,"checkerConformance","Document Checker Conformance")%>This section uses the words MUST, MUST NOT, SHOULD and MAY as in <%=ref("RFC 2119")%>.
<%=head(3,"syntaxChecker","Syntax Checker")%>
An OWL
syntax checker
takes a document as input, and returns one word being one of Lite
,
DL
, Full
, Other
.
The return value MUST conform with the following:
In addition, an OWL Syntax Checker SHOULD report a warning if
the
RDF graph
<%=ref("RDF Concepts")%>
corresponding to the document
uses any URI references
starting with the prefix http://www.w3.org/2002/07/owl#
except those found in the
<%=ref("RDF Schema for OWL")%>.
An OWL syntax checker SHOULD report network errors occurring during the computation of the imports closure.
<%=head(3,"consistencyChecker","Consistency Checker")%>
An OWL consistency checker
takes a document as input, and returns one word being Consistent
,
Inconsistent
, or Unknown
.
An OWL consistency checker SHOULD report network errors occurring during the computation of the imports closure.
An OWL consistency checker MUST provide a means to determine the datatypes supported by its datatype map, <%=ref("OWL Semantics and Abstract Syntax")%>; for example, by listing them in its supporting documentation.
An OWL consistency checker MUST provide a means to determine the model theory <%=ref("OWL Semantics and Abstract Syntax")%>, it uses (either the Direct Model-Theoretic Semantics or the RDF-Compatible Model-Theoretic Semantics); for example, in its supporting documentation.
An OWL
consistency checker MUST be sound:
it MUST
return Consistent
only when the
input document is consistent and Inconsistent
only when the input
document is not consistent, with respect to the datatype map of the checker.
If an input document uses datatypes that are not supported by the datatype map of an OWL consistency checker then it MAY report a warning.
An OWL consistency checker is
complete and terminating,
if, given sufficient (but
finite) resources (CPU cycles and memory)
and the absence of
network errors, it will always return
either Consistent
or Inconsistent
. It has
been shown that for OWL Lite and DL it is possible to construct a
complete and terminating consistency checker
(the languages are decidable),
and that
for OWL full it is not possible to construct a complete and terminating
consistency
checker (the language is undecidable,
<%=ref("Practical Reasoning")%>).
The
datatype map of
an OWL consistency checker MUST minimally support at least
xsd:integer
, xsd:string
from <%=ref("XML Schema Datatypes")%>.
An OWL consistency checker SHOULD NOT return
Unknown
.
Unknown
, while sometimes needed, is not
a desired response.
Four different conformance classes of OWL consistency checker are defined.
An OWL Lite consistency checker is an OWL consistency checker that takes an OWL Lite document as input, and uses the Direct Model-Theoretic Semantics.
An OWL DL consistency checker is an OWL consistency checker that takes an OWL DL document as input and uses the Direct Model-Theoretic Semantics.
An OWL Full consistency checker is an OWL consistency checker that takes an OWL Full document as input and uses the RDF-Compatible Model-Theoretic Semantics.
The
datatype map of an
OWL Full consistency checker
MUST also support
rdf:XMLLiteral
from <%=ref("RDF Concepts")%>,
see <%=ref("RDF Semantics")%>.
A complete OWL Lite consistency checker is an OWL Lite consistency checker that is complete and terminating.
Note: An OWL Full consistency checker may indicate that an OWL DL document is inconsistent, while an OWL DL consistency checker indicates that the same document is consistent, (for example: compare test Thing-005 with Thing-004 or compare AnnotationProperty-001 with AnnotationProperty-002). Every OWL DL consistency checker is also an OWL Lite consistency checker.
A complete and terminating OWL DL consistency checker is an OWL consistency checker that is complete and terminating.
--%>Note:
<%--
decisive OWL DL consistency checkers
and --%>
A
complete OWL Lite consistency checker
MAY return Unknown
for an OWL Lite document in the case where
a resource limit has been exceeded.
Note: The usage of the word 'complete' in this section follows the conventions of the description logic community. In some other communities the word 'complete' is used in a weaker sense, refering to the detection of inconsistency by logical inference systems.
An OWL syntax checker when presented with any of the test files must return the indicated result. <%----%> This includes the extra credit tests. <%----%>
<%=head(2,"runningConsistencyChecker","OWL Consistency Checker")%>An OWL consistency checker can be tested using appropriate consistency and inconsistency tests. Appropriate tests are those of an appropriate level and for which the checker has appropriate datatype support. The level of the test indicates the semantic theory being used, which may differ from the level of the file. For example, test Thing-004 contains an OWL DL file which is consistent as an OWL DL consistency test, but inconsistent as an OWL Full consistency test.
An OWL consistency checker has appropriate datatype support for a test if both:
An OWL Lite consistency checker
with
appropriate datatype support,
when presented with a file from
an OWL Lite consistency test,
must return Consistent
or Unknown
.
An OWL DL consistency checker
with
appropriate datatype support,
when presented with a file from
an OWL DL or OWL Lite consistency test,
must return Consistent
or Unknown
.
An OWL Full consistency checker
with
appropriate datatype support,
when presented with a file from
an OWL Full consistency test,
must return Consistent
or Unknown
.
The corresponding inconsistency tests must return
Inconsistent
or Unknown
.
A complete OWL Lite consistency checker
<%--or a
complete and terminating OWL DL consistency
checker --%> should not return Unknown
on the OWL Lite
consistency
or inconsistency tests, regardless of the use of
unsupported datatypes.
The above constraints also apply to
extra credit tests.
Consistency checkers that return the correct answer (i.e. not Unknown
)
gain the extra credit.
The Manifest
file follows the RDF schema developed
for the RDF Test Cases <%=ref("RDF Tests")%>.
This is augmented by a few new properties and types which are declared in the OWL Test Ontology, found at http://www.w3.org/2002/03owlt/testOntology.
Specifically each test has its own Manifest
file, and is identified from
the URI reference formed from the Manifest
file's URL with a fragment test
.
The test has one rdf:type
explicit, and this is one of:
otest:NotOwlFeatureTest
otest:PositiveEntailmentTest
otest:NegativeEntailmentTest
otest:TrueTest
otest:OWLforOWLTest
otest:ConsistencyTest
otest:InconsistencyTest
otest:ImportEntailmentTest
otest:ImportLevelTest
Where otest
is bound to
http://www.w3.org/2002/03owlt/testOntology#
and rtest
is bound to
http://www.w3.org/2000/10/rdf-tests/rdfcore/testSchema#
.
The name of the original author of the test is shown using a
dc:creator
property, see <%=ref("Dublin Core")%>.
A description of the test is given (using XHTML markup <%=ref("XHTML")%>)
as the value of the rtest:description
property.
An issue, if any, from the OWL Issues list <%=ref("OWL Issues")%>, is
the value of a rtest:issue
property.
An appropriate language feature, from the OWL namespace, if any, is
the value of the otest:feature
property.
The input documents with the test data are found as the value of
the rtest:inputDocument
property or
as the value of both the
rtest:premiseDocument
and
the
rtest:conclusionDocument
.
The support files for import entailment tests, import level tests
and test imports-002 are found
as the values of otest:importedPremiseDocument
.
The conformance levels associated with both files and tests
are given with the otest:level
property.
The value for each document is one of
otest:Full
, otest:DL
,
otest:Lite
or otest:Other
.
Each test is explicitly associated with one or two levels.
If it is associated with otest:Lite
then it
is implicitly suitable for otest:DL
.
The datatypes used in the test are given with the
otest:usedDatatype
property or with one of its subproperties:
otest:supportedDatatype
or otest:notSupportedDatatype
.
These
indicate that
the test is only valid when the datatype is supported or not supported respectively
by the
datatype map being used.
The rtest:status
of the test
reflects the process of
appendix A.
It
is given as one of the following levels:
The existence of an issue in this section does not constitute a promise that this will be addressed before the Proposed Recommendation.
<%=head(2,"editorialIssuesMachine","Arising from Automated Checking")%>Most of these issues will be ignored.
<%=approved.getErrors()%> <%=Manifest.loader.getErrors()%>
<% } %>This index does not differentiate between approved and proposed tests.
--%> <%=head(2,"testIndexFunction","Index of OWL Feature Tests")%> <% subIndex = featureIndex; %> <%@ include file="subIndex.jsp" %> <%=head(2,"testIndexIssue","Index of OWL Issue Tests")%> <% subIndex = issueIndex; %> <%@ include file="subIndex.jsp" %> <%=head(2,"testIndexMisc","Index of Miscellaneous Tests")%> <% subIndex = miscIndex; %> <%@ include file="subIndex.jsp" %> <%=head(2,"testIndexDL","Index of Description Logic Tests")%> <% subIndex = dlIndex; %> <%@ include file="subIndex.jsp" %> <%=head(2,"testIndexExtra","Index of Extra Credit Tests")%> <% subIndex = extraIndex; %> <%@ include file="subIndex.jsp" %> <% } %>Jeremy Carroll thanks Oreste Signore, his host at the W3C Office in Italy and Istituto di Scienza e Tecnologie dell'Informazione "Alessandro Faedo", part of the Consiglio Nazionale delle Ricerche, where Jeremy is a visiting researcher.
The following people have contributed tests to this document:
<% { String a[] = Manifest.authors();
for (int i=0;i
Ian Horrocks contributed to the conformance section of this document.
Sandro Hawke created the tests results page, that has been a great help during the Candidate Recommendation phase.
We thank those who gave test reports and other feedback during the Candidate Recommendation: Ken Baclawski, Sean Bechhofer, Ian Dickinson, Michael Grove, Sandro Hawke, Ian Horrocks, Minsu Jang, Gary Ng, Mehrdad Omidvari, Bijan Parsia, Peter F. Patel-Schneider, Dave Reynolds, Rob Shearer, Evren Sirin, Charles White and Youyong Zou. We also thank the many others who helped develop the systems which produced these reports.
This document is the result of extensive discussions within the Web Ontology Working Group as a whole. The partipants in this Working Group included: Yasser alSafadi, Jean-François Baget, James Barnette, Sean Bechhofer, Jonathan Borden, Stephen Buswell, Jeremy Carroll, Dan Connolly, Peter Crowther, Jonathan Dale, Jos De Roo, David De Roure, Mike Dean, Larry Eshelman, Jérôme Euzenat, Tim Finin, Nicholas Gibbins, Sandro Hawke, Patrick Hayes, Jeff Heflin, Ziv Hellman, James Hendler, Bernard Horan, Masahiro Hori, Ian Horrocks, Jane Hunter, Rüdiger Klein, Natasha Kravtsova, Ora Lassila, Deborah McGuinness, Enrico Motta, Leo Obrst, Mehrdad Omidvari, Martin Pike, Marwan Sabbouh, Guus Schreiber, Noboru Shimizu, Michael K. Smith, John Stanton, Lynn Andrea Stein, Herman ter Horst, David Trastour, Frank van Harmelen, Bernard Vatant, Raphael Volz, Evan Wallace, Christopher Welty, Charles White, Frederik Brysse, Francesco Iannuzzelli, Massimo Marchiori, Michael Sintek and John Yanosy.
<%=head(1,"changes-since-PR","Changes Since Proposed Recommendation")%>This section gives the changes between this document and the OWL Test Cases Proposed Recommendation.
The term datatype map is used instead of the term datatype theory, for consistency with the OWL and RDF Semantics. This occurred a number of times, including in the descriptions of tests miscellaneous-204, miscellaneous-205 and I5.8-012. The last of these consequentially required other minor rephrasing.
Updated references to RDF and OWL documents.
Added a paragraph near end of section 4.2.2, clarifying that a datatype map of an OWL Full consistency checker, (being a datatype map from RDF Semantics) "MUST" contain an entry for rdf:XMLLiteral. This makes explicit a requirement that was already implicit in the PR document. Also clarified that the datatype map in the definition of an OWL Full consistent document is as defined in RDF Semantics, by changing the link.
Consequentially, made explicit reference to RDF Semantics (this reference was implicit in the OWL Test Proposed Recommendation).
Corrected an error in the metadata of test miscellaneous-205 which is not applicable for OWL Full, since rdf:XMLLiteral is a required datatype for OWL Full. This change is visible as the deletion of the word "Full" from the header of the test.
This error in the OWL Test Candidate and Proposed Recommendation appears to have been relatively benign: