Re: [ontolog-forum] Are there any plans to develop an OWL version of QUD

[Kingsley Idehen <kidehen@xxxxxxxxxxxxxx>]

On 5/8/14 9:35 PM, Simon Spero wrote:

[How did I end up defending OWL 2 DL when my real problem is with vocabularies be described as OWL that aren't even consistent] 

Kingsley - 

   the first  issue is that the datatype I used as an example is defined using OWL, in the very same document.  The lexical space of datatypes defined in OWL is empty; any literal that uses an OWL defined type as the datatype is ill-typed (the type is not unknown to the processor).   Even if it weren't blocked by RDF-1.1, it is explicitly forbidden by the OWL 2 specification. 

It seems that the tools in question do not do any  validation of literal types, even for predefined types which are in conflict with the specified rdfs:range.       

After defining a bunch of ontology metadata properties as data properties rather than annotation properties explicitly so that they can define axioms on them, they go on to  use those properties with a literal incompatible with its range.    

If one says one is  doing  OWL full, do OWL full;   if  OWL-DL, do OWL-DL; for RDFS, or RDF, do RDFS, or RDF.  

But don't say that you're doing something, do something different and broken, then complain.   

An example of doing things the right way is the Virtuoso sparql service description. Although Virtuoso supports most of SPARQL 1.1Query, because some parts aren't there, the service description only claims to support 1.0. 

Because none of the entailment regimes match the supported entailments, none are claimed, even though role and class inheritance can be selected. 

 (also, much  the entailment regime document doesn't make sense any more with RDF 1.1).

Simon

Simon,

I yanked out your thread from Schema.org, but it wasn't to imply you were defending OWL 2 DL or anything like that, I just wanted to shed more light on the fact that OWL isn't a monolith, and the fact that reasoning using terms from the owl: namespace doesn't automatically imply full support and endorsement of everything associated with OWL and its history.

For instance, when I demonstrate inference and reasoning it typically boils down to demonstrating practical use of one or more of the following:

1. owl:sameAs
2. owl:InverseFunctionalProperty
3. owl:SymmetricProperty
4. owl:equivalentClass
5. owl:equivalentProperty
6. owl:inverseOf .

In a sense, I was also trying to clarify a few issues with John Sowa (and others on this list) in regards OWL and its practical use.

Kingsley

On Thu, May 8, 2014 at 7:59 AM, Kingsley Idehen <kidehen@xxxxxxxxxxxxxx> wrote:

All,

I am sharing this thread from the Schema.org forum so at to shed light on confusion that tends to arise in regards to RDF (language for creating structured data with varying degrees human and machine comprehensible entity relations semantics) and OWL.

John Sowa:

When I make reference (for instance) to practical applications (e.g., inference and reasoning in regards to heterogeneous data integration etc.) of the Semantic Web via the Linked Open Data cloud, its very much in line with the worldview expressed by Holger.

Please read on...

On 5/8/14 1:31 AM, Holger Knublauch wrote:

Simon,

On 5/7/2014 6:07, Simon Spero wrote:

There are a number of reasons why the vocabularies at http://qudt.org/ are not legal OWL or RDF. 

They are perfectly legal RDF, but indeed not legal OWL DL. As stated elsewhere, not being OWL DL compliant is not necessarily a bad thing.

Issue 1:  You can’t use user defined datatypes as the type of a literal 

Yes in OWL, but in RDF this is perfectly valid, see

http://www.w3.org/TR/2014/REC-rdf11-concepts-20140225/#dfn-recognized-datatype-iris

"Semantic extensions of RDF might choose to recognize other datatype IRIs..."

In the case of QUDT I would even argue that it is of great value to create literals such as "4.2"^^qudt:Meter that are a much more meaningful information exchange format than "4.2"^^xsd:decimal. Of course OWL DL would prohibit this, but OWL DL tools can probably run such literals through a normalizer algorithm that also converts feet to meters etc. Oh but I forgot... OWL cannot even be used to check if a > b or a + b so nevermind.

There are no reasoners for OWL 2 Full.

Not true. For example the TopSPIN engine (part of TopBraid Composer Free Edition) can run rule on any RDF graph, even with user-defined datatypes etc. It's just a matter of writing the right rules. Unless of course you mean "Reasoner" = "OWL DL Reasoner", which is not my understanding. On a general discussion, see

http://composing-the-semantic-web.blogspot.com.au/2010/04/where-owl-fails.html

Corollary 3: The QUDT ontologies cannot be edited or processed using OWL tooling.

Not true, they were edited with TopBraid Composer, which is an OWL tool (and also an RDF tool).

Why this matters

The RDF to OWL parsers in OWLAPI (used by tools like Protegé) will try their best

Proper RDF based tools based on APIs such as Jena or Sesame have no problems with these files. You keep coming back to OWL DL compliance, and I keep coming back to the fact that it was an (unfortunate) design decision by the OWL 2 community and the OWL API to abandon the RDF roots. If these design decisions now fire back into RDF and OWL people want to enforce their world view upon RDF graphs only because their tools cannot handle RDF, then we need to decisively make clear what the source of the issue is.

And to be very clear: the OWL DL community has way too long dominated ontology development. There is already sufficient value in the basic principles of RDF such as URIs and the ability to link anything with everything, and the adoption of these simple principles is hindered by an overly complex layer cake of features and constraints that few people really need. A generation of graduates leaves their universities with the impression that the semantic web is OWL DL, and early adopters of this technology in industry are overwhelmed by the complexity of the whole stack. RDF and SPARQL is really all that is needed to know, and with SPARQL's CONSTRUCT you even get a rule language for free (see for example http://spinrdf.org).

(3) The RDF Semantics specification (hereafter RDF-MT) §7 states that:

          "[a]n ill-typed literal is one whose datatype IRI is recognized, but whose character string is assigned no value by the lexical-to-value mapping for that datatype."

(4) RDF-MT §7.1 states that:

"any triple, and hence any graph, containing an ill-typed literal will be D-unsatisfiable, i.e. false in every D-interpretation. This applies only to literals typed with recognized datatype IRIs in D".

(5) RDF-MT §7.2 notes that a D-unsatisfiable graph entails any other graph by the principle of ex falso quodlibet [xkcd]. 

For more details see e.g. the discussion on the principle of non-contraction in the Stanford Encyclopedia of Philosophy: Classical Logic, §3

The VAEM document at http://www.linkedmodel.org/1.2/schema/vaem includes the following triples:

(6) http://www.linkedmodel.org/1.2/schema/vaem a owl:Ontology . 

(7) vaem:dateUnion a rdfs:Datatype . 

(8) http://www.linkedmodel.org/1.2/schema/vaem vaem:dateCreated "2011-04-20"^^vaem:dateUnion .       

Proof 

(9)       http://www.linkedmodel.org/1.2/schema/vaem is declared to be an owl ontology  (2b,6).

(10)     The IRI vaem:dateUnion is declared as a datatype and hence is known(1,2a,7).

(11)     Any RDF literal with the datatype IRI  vaem:dateUnion is ill typed (1,3,10).

(12)     The graph at  http://www.linkedmodel.org/1.2/schema/vaem is D-unsatisfiable  (4,8,11). 

(13)     The graph at  http://www.linkedmodel.org/1.2/schema/vaem entails any other RDF graph (5,12) □

I do not believe these statements are correct. "2011-04-20"^^vaem:dateUnion is a perfectly valid RDF literal.

If a problem falls within the class of problems that OWL 2 DL can handle, it is good to stay within the constraints of DL. Going to OWL Full should be made on a carefully considered basis of costs vs. benefits.

This is a theory, not a fact. One could argue the other way around: if a problem can be solved in RDF Schema, then why use OWL?

Thirdly: The vocabularies being discussed were inconsistent at every level of the RDF, RDFS, and OWL chain, and contained basic type errors.

I don't see where they were invalid RDF, please clarify.

I do not know what tools were used to create them, but whatever tool it was could not do basic consistency checking.

They were created with TopBraid and TopBraid uses Jena which has no problems with those files either. Only because they are not valid OWL DL does not mean they are inconsistent.

Holger

-- 

Regards,

Kingsley Idehen	      
Founder & CEO 
OpenLink Software     
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter Profile: https://twitter.com/kidehen
Google+ Profile: https://plus.google.com/+KingsleyIdehen/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen

_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
 

 
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
 

-- 

Regards,

Kingsley Idehen	      
Founder & CEO 
OpenLink Software     
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter Profile: https://twitter.com/kidehen
Google+ Profile: https://plus.google.com/+KingsleyIdehen/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen

smime.p7s
Description: S/MIME Cryptographic Signature

_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (01)