- From: Young,Jeff (OR) <jyoung@oclc.org>
- Date: Thu, 8 Jul 2010 16:57:10 -0400
- To: "Bernard Vatant" <bernard.vatant@mondeca.com>, "public-lld" <public-lld@w3.org>
- Message-ID: <52E301F960B30049ADEFBCCF1CCAEF5908E8F5A8@OAEXCH4SERVER.oa.oclc.org>
Bernard,
Your two quick points raise some in my mind:
· I looked at the first example of "Correct Usages" in the "Overloading" document but couldn't make heads or tails of it with live information. If someone could rationalize the claim of "correct usage" in this example, I might take the document more seriously:
o http://ontologydesignpatterns.org/wiki/Community:Overloading_OWL_sameAs#Summary_and_Synthesis
· I agree that owl:sameAs assertions found in the wild will never be trustworthy. I assume we all agree, though, that institutions like libraries should conform to Web standards and thus remain relevant.
· IMO, understanding how to consume Linked Data effectively isn't as important as understanding how to produce it clearly and efficiently from information we already have.
· My amazing argument against multiple rdf:types is more a case of being misunderstood (my own fault). I'm really not in denial of rdf:type implications of rdfs:subClassOf and rdfs:Property.
· I think your last paragraph hit the nail on the head and bears repeating:
"That said, in an open world, an application will be able to pick in the descriptions the elements it can consume. If something (someone) is declared with rdf:type foaf:Person and entailed some way to be also skos:Concept, if my application is interested in the social aspects of the description for a social web applications, I will consider only the triples with predicates in the FOAF namespace, and if your application is interested only by this resource as an entry in a resource index, using e.g. dcterms:subject or dcterms:creator, you will pick only the predicates in the SKOS namespace (prefLabel, altLabel ...)"
That pretty much captures the future as I see it too. How would you feel about these as a set of URI patterns to help libraries bridge the gap from records to resources:
http://example.org/marc/12345 (303 redirect to...)
http://example.org/marc/12345/ (generic resource content-negotiable to...)
http://example.org/marc/12345/default.html
http://example.org/marc/12345/default.fr.html
http://example.org/marc/12345/default.en.html
http://example.org/marc/12345/default.gr.html
http://example.org/marc/12345/marc.html
http://example.org/marc/12345/marc.xml
http://example.org/marc/12345/frbr.html
http://example.org/marc/12345/frbr.rdf
http://example.org/marc/12345/skos.html
http://example.org/marc/12345/skos.rdf
http://example.org/marc/12345/all.rdf
etc.
There are obviously other types of records besides MARC, but the general pattern should hold. Do you see your point lurking in there somewhere?
Jeff
From: public-lld-request@w3.org [mailto:public-lld-request@w3.org] On Behalf Of Bernard Vatant
Sent: Thursday, July 08, 2010 5:34 AM
To: public-lld
Subject: Re: MARC Codes for Forms of Musical Composition
Hello all
Two quick points
1. Overloading or abuse of owl:sameAs in linked data land is a well-known issue that has been discussed at length, before and beyond the (excellent) quoted paper. A good account of the debate can be found at
http://ontologydesignpatterns.org/wiki/Community:Overloading_OWL_sameAs
Identity is a tricky issue which is representative of the discrepancy between the "hard" semantics as declared by standards, and the not-so-hard and various ones implicitly understood by the users, who tend to hack the original semantics, either because they do not read the specs, or misunderstand them, or use classes and properties not exactly meaning what they want, default of more precise ones. Use and abuse of owl:sameAs is typivcal of this. It's pretty clear that buying all owl:sameAs links in the linked data cloud to mean what the OWL specification says it means will entail zillions of inconsistencies of all kinds, the most obvious being that things considered as distinct here will be merge there. There is no way to bring global consistency to this "knowledge soup", what is needed is ways to sort it through various heuristics.
2. I'm amazed that one would debate about unicity of rdf:type at all. It's certainly a good practice for the URI publisher to declare a single rdf:type. But based on OWL or RDFS semantics, other types will be entailed even if they are not declared.
for example, to keep it simple :
:x rdf:type foaf:Person will entail
:x rdf:type foaf:Agent based on foaf:Person rdfs:subclassOf foaf:Agent
other types will be entailed from domain declarations of properties used in the description etc.
In the open world where the URI is used and re-used, linked to and from, it's obvious that new types will be acquired by entailments. And to be back to point 1, in particular applying strictly owl:sameAs semantics will bring about a bunch of possibly conflicting types.
That said, in an open world, an application will be able to pick in the descriptions the elements it can consume. If something (someone) is declared with rdf:type foaf:Person and entailed some way to be also skos:Concept, if my application is interested in the social aspects of the description for a social web applications, I will consider only the triples with predicates in the FOAF namespace, and if your application is interested only by this resource as an entry in a resource index, using e.g. dcterms:subject or dcterms:creator, you will pick only the predicates in the SKOS namespace (prefLabel, altLabel ...)
Bernard
2010/7/6 Young,Jeff (OR) <jyoung@oclc.org>
Let me address Ross' question before attempting to argue that restraint to a single rdf:type is good practice.
Here is the example in question:
http://purl.org/NET/marccodes/muscomp/sy.rdf
The owl:sameAs property asserts that these two URIs identify "the same thing" (http://www.w3.org/TR/owl-ref/#sameAs-def):
http://purl.org/NET/marccodes/muscomp/sy#genre
http://dbpedia.org/resource/Symphony
The 1st URI responds with this statement:
<http://purl.org/NET/marccodes/muscomp/sy#genre> rdf:type <http://purl.org/ontology/mo/Genre>
The 2nd URI responds with this:
<http://dbpedia.org/resource/Symphony> rdf:type <http://sw.opencyc.org/2008/06/10/concept/Mx4rwSmVfJwpEbGdrcN5Y29ycA>
<http://dbpedia.org/resource/Symphony> rdf:type <http://sw.opencyc.org/2008/06/10/concept/Mx4rvcNktpwpEbGdrcN5Y29ycA>
Other rdf:type and owl:sameAs assertions cascade from there in dbpedia.
The following document isn't authoritative, but it discusses some of the confusion surrounding owl:sameAs and may also help us sort out the issues:
http://www.w3.org/2009/12/rdf-ws/papers/ws21.
Here is a quote:
"However, owl:sameAs does have a particular semantics of individual identity, namely that the two individuals are exactly the same and so share all the same properties." (original emphasis).
Since rdf:type is a property, I assume that an OWL reasoner should back me up in my claim that Ross' example has multiple rdf:types. I just downloaded Pellet and will report on the results once I figure out how to run it. Hopefully, it will demonstrate how "share" involving owl:sameAs plays out in practice.
Jeff
From: rxs@talisplatform.com [mailto:rxs@talisplatform.com] On Behalf Of Ross Singer
Sent: Monday, July 05, 2010 10:03 PM
To: William Waites
Cc: Young,Jeff (OR); Antoine Isaac; Karen Coyle; public-xg-lld@w3.org; List for Working Group on Open Bibliographic Data; public-lld
Subject: Re: MARC Codes for Forms of Musical Composition
My question was more based on the fact that I don't think anything should have explicitly set multiple rdf:types in there.
If so, I'm curious to what they are.
-Ross.
On Mon, Jul 5, 2010 at 3:35 PM, William Waites <william.waites@okfn.org> wrote:
On 10-07-05 10:35, Ross Singer wrote:
> Jeff, which resources have multiple rdf:types? Of the muscomps, they
> should all only be mo:Genre.
I think it is perfectly valid to have multiple types. At the
very minimum everything is an rdfs:Resource whether
stated explicitly or not. If something breaks when it is
explicitly stated because it doesn't like multiple types I
think that something is itself broken...
Cheers,
-w
--
William Waites <william.waites@okfn.org>
Mob: +44 789 798 9965 Open Knowledge Foundation
Fax: +44 131 464 4948 Edinburgh, UK
RDF Indexing, Clustering and Inferencing in Python
http://ordf.org/
--
Bernard Vatant
Senior Consultant
Vocabulary & Data Engineering
Tel: +33 (0) 971 488 459
Mail: bernard.vatant@mondeca.com
----------------------------------------------------
Mondeca
3, cité Nollez 75018 Paris France
Web: http://www.mondeca.com
Blog: http://mondeca.wordpress.com
----------------------------------------------------
Received on Thursday, 8 July 2010 20:58:17 UTC