W3C home > Mailing lists > Public > public-lld@w3.org > July 2010

RE: [open-bibliography] MARC Codes for Forms of Musical Composition

From: Houghton,Andrew <houghtoa@oclc.org>
Date: Wed, 7 Jul 2010 16:30:08 -0400
Message-ID: <6548F17059905B48B2A6F28CE3692BAA07589ED1@OAEXCH4SERVER.oa.oclc.org>
To: "public-lld" <public-lld@w3.org>
Cc: "Ross Singer" <ross.singer@talis.com>, "Young,Jeff (OR)" <jyoung@oclc.org>
> From: public-lld-request@w3.org [mailto:public-lld-request@w3.org] On Behalf Of Ross Singer
> Sent: Wednesday, July 07, 2010 02:55 PM
> To: Young,Jeff (OR)
> Cc: Erik Hetzner; public-lld
> Subject: Re: [open-bibliography] MARC Codes for Forms of Musical Composition

>I guess what I'm saying is that this viewpoint has no basis in the common practice of how linked data is being created currently.

With respect, that is not an argument...

Just because it has been done forever, since the dawn of time,
doesn't mean that it is correct:

  Witness Apple's iPhone signal strength bars... works great on 
  iPhone, iPhone 3G, iPhone 3Gs, oops... iPhone 4G its been wrong 
  since the dawn of time...

DBPedia is a poor example to use for common practice. It certainly
is a large dataset, but only one dataset. Also, there has been 
numerous discussions on the SKOS list over the past year about
DBPedia's describing resources with multiple rdf:type is not 
considered "good" practice.

Jeff Young and I were just talking a bit about issues around 
conflation of individuals. What if you had a person and music 
band by the same name because the person is a solo singer. So
you describe the person as foaf:Person and the band as 
foaf:Group or foaf:Organization. Since foaf: does not include 
a property to specify birth date you decide to use the bio 
ontology to supplement the foaf: ontology:

<foaf:Person rdf:about="#Me">
  <bio:Birth rdf:parseType="Resource">
    <dct:date>2000-01-01</dct:date>
  </bio:Birth>
</foaf:Person>

<foaf:Group rdf:about="#Me" />

Because the triples have the same subject: is bio:Birth
associated with the foaf:Group or the foaf:Person? Doesn't
make any sense that a band has a birth date does it? Maybe
a startup/disband date. This is the sort of nonsense you 
wind up with when you conflate individuals and you use 
properties declared outside the ontology described by 
rdf:type.

Things can even get complex for properties in the ontology 
specified by rdf:type. An OWL reasoner could deduce that a 
property with a domain foaf:Person can only be associated 
with the foaf:Person description, but what about a property 
that has multiple domains? Which class in the conflation 
does it belong to? When you conflate individuals, you start 
having problems making sense of what is being said about 
what.

BTW, sure I can intuit that bio:Birth should go with the
foaf:Person, but can a machine do that for any unknown
ontology... most likely not.


Andy.
Received on Wednesday, 7 July 2010 20:31:20 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:27:36 UTC