- From: Young,Jeff (OR) <jyoung@oclc.org>
- Date: Thu, 8 Jul 2010 09:36:09 -0400
- To: "Erik Hetzner" <egh@e6h.org>, "public-lld" <public-lld@w3.org>
We don't need to call this a rule or best practice. Call it an appeal to respect subtle differences. Here's Ross' example taken to its logical conclusion: <http://purl.org/NET/book/isbn/0192838024#book> a <http://purl.org/NET/book/vocab#Book>, <http://purl.org/ontology/bibo/Book>, <http://vocab.org/frbr/core#Work> , <http://vocab.org/frbr/core#Expression> , <http://vocab.org/frbr/core#Manifestation> , <http://vocab.org/frbr/core#Item> . Here's a quote from Barbara Tillett's "What is FRBR?" <http://www.loc.gov/cds/downloads/FRBR.PDF> "Before FRBR our cataloging rules tended to be very unclear about using the words “work,” “edition,” or “item.”2 Even in everyday language, we tend to say a “book” when we may actually mean several things. Jeff -----Original Message----- From: public-lld-request@w3.org [mailto:public-lld-request@w3.org] On Behalf Of Erik Hetzner Sent: Thursday, July 08, 2010 1:03 AM To: public-lld Subject: Re: [open-bibliography] MARC Codes for Forms of Musical Composition At Wed, 7 Jul 2010 16:31:07 -0400, Young,Jeff (OR) wrote: > > Erik, > > What Andy and I are suggesting are not "arbitrarily different URIs". > These are different URIs that preserve (rather than conflate) the > identity of things with different types (especially across different > ontologies). Ross said it as well as anyone possibly could: "Other > people think differently". Our suggestion is to accept and respect these > differences. Hi Jeff, I find myself agreeing with your arguments, but not with the conclusion. The arguments that I have heard do not justify defining a rule or even best practice that multiple rdf:types should not be assigned to a single resource. I have not heard an answer to Ross’ question to what is wrong with: <http://purl.org/NET/book/isbn/0192838024#book> a <http://purl.org/NET/book/vocab#Book>, <http://purl.org/ontology/bibo/Book>, <http://vocab.org/frbr/core#Manifestation> . If I imagine the alternative: <http://purl.org/NET/book/isbn/0192838024#book> a <http://purl.org/NET/book/vocab#Book> . <http://purl.org/NET/book/isbn/0192838024#bibo> a <http://purl.org/ontology/bibo/Book> . <http://purl.org/NET/book/isbn/0192838024#frbr> a <http://vocab.org/frbr/core#Manifestation> . I an think of at least the following difficulties, off the top of my head: a) Using rdf:seeAlso between these URIs does not adequately describe their relationship. This could obviously be addressed. b) If one person links to #bibo and one to #frbr, they need more information to know that they are talking about the same thing. c) Even having this information, any queries, etc. against the data will involve two steps, e.g.: SELECT * WHERE { ?b1 ex:hasVariant ?b2 . ?b1 rdf:type vocab:Book . ?b2 rdf:type frbr:Manifestation . ?p1 ex:likes ?b1 . ?p2 ex:likes ?b2 . } > Splitting hairs on "person" misses the point that the interpretation of > similarities and differences different ontologies is highly subjective. > Freely conflating types on a subjective basis undermines unexpected > reuse where the distinctions are important. Likewise, requiring every resource to have a single rdf:type hierarchy undermines expected reuse when the distinctions are unimportant. best, Erik
Received on Thursday, 8 July 2010 13:37:06 UTC