W3C home > Mailing lists > Public > public-semweb-lifesci@w3.org > March 2013

Re: owl:sameAs - Harmful to provenance?

From: Michael Erdmann <erdmann@diqa-pm.com>
Date: Thu, 28 Mar 2013 09:31:47 +0100
Message-ID: <5153FFF3.90501@diqa-pm.com>
To: public-semweb-lifesci@w3.org
On 28.03.2013 06:18, Jeremy J Carroll wrote:
> I rather liked the GOTO/owl:sameAs presentation …
> and I am not sure that the problem can be explained away as modeling errors.
> To me it does seem that there is a case to answer … modeling is necessarily approximate and  it is not realistic to expect equality up to some approximation to be an equivalence relation.
> Yet the Semantic Web is about linking together many somewhat approximate models.
Agreed, but owl:sameAs, has a very strict semantics and cannot be 
mis-interpreted. There are other, better fitting standardized properties 
which have fewer implications but nevertheless, are grounded in the OWL 
semantics. The SKOS specification introduces mapping properties to align 
concepts [1]. The spec says:

    "The property skos:exactMatch is used to link two concepts,
    indicating a high degree of confidence that the concepts can be used
    interchangeably across a wide range of information retrieval
    applications. skos:exactMatch is a transitive property ... and an
    instance of owl:SymmetricProperty"

The effect of A owl:sameAs B is, that A and B become two names for the 
same thing, thus statements about A and B are merged and provenance 
information is lost. After stating A skos:exactMatch B (and also A 
skos:closeMatch B, for that matter) there are still two things with 
statements about them coming from different sources.

Thus, there is nothing wrong with the semantics of owl:sameAs, only in 
the way it is used. We should try to use more appropriate language in 
our RDF models, in order for application developers to create useful 
applications that interpret the models in a standards-conform way. In 
the end, that is why we have (and want) a formally defined semantics of 
a language.

> I also rather liked the title because I used a similar title on the 9th of November, 1989, at a presentation in Saarbrücken, Germany. My title was [Graph] "unification considered harmful". Given the date and location it was the least well chosen title I have ever used.

michael (from Germany ;)

[1] http://www.w3.org/TR/skos-reference/#mapping
> Jeremy J Carroll
> Principal Architect
> Syapse, Inc.

Dr. Michael Erdmann   |    erdmann@diqa-pm.com    |   +49 151 6140 1790

DIQA Projektmanagement GmbH | Pfinztalstr 90 | 76227 Karlsruhe | Germany
Handelsregister: Amtsgericht Mannheim HRB 715454 | USt-IdNr: DE283037270
Geschäftsführer: Dr. Michael Erdmann, Dipl.-Wirtsch.-Inf. Daniel Hansch

This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.
Received on Thursday, 28 March 2013 13:17:11 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:53:02 UTC