On Tue, Apr 13, 2010 at 11:15 AM, Karen Coyle <lists_at_kcoyle.net> wrote:
>> For example, they link to:
>> http://dbpedia.org/page/Wikipedia:WikiProject_Bob_Dylan
>>
>> which is a *page* about the representation of Bob Dylan (not the
>> representation of Bob Dylan himself)
>
> I must say that I don't see a significant difference between the use of a
> wikipedia page or the use of an LC name authority record to represent Dylan
> -- since Dylan himself cannot be reduced to linked data. I think we are all
> more comfortable with the LCNA because we understand its function as an
> identifier, but it's just a record in the same sense that wikipedia is just
> a page.
Yes, well, it's not like the linked data community hasn't thought of
this or addressed this in some way (and I totally admit it's the
Kool-Aid of linked data -- you either take it or you don't) - it's the
distinction between information resources and non-information
resources (brief into here:
http://www.rdfabout.com/intro/?section=8).
Bob Dylan as an "entity" (that is, an identifier that represents Bob
Dylan) *can* be expressed, whether we choose to use the form
http://dbpedia.org/resource/Bob_Dylan or info:lccn/n50030190. These
represent the identity "Bob Dylan" and we can make assertions about
him. The wikipedia page (and, by extension, the DBpedia 'page' URI
that VIAF points to) are documents about Bob Dylan -- so if you refer
to those URIs, you're talking about the page about Bob Dylan, not the
singer-songwriter himself. This level of indirection is important -
it helps differentiate documents from data.
I'm not saying that the Wikipedia URI isn't totally useful (pretty
easy hop to dbpedia, after all), it's just the way that OCLC is using
it is wrong, in this case.
>
>> -- they also model all of their
>> identities as skos:Concepts, which is problematic in other ways.
>
> Yes, SKOS gets used because it's Simple and it is There. We need to develop
> some other "S"s - simple formats for other kinds of data.
>
Actually, SKOS as "simple" isn't the problem. The problem is if you
think there's a difference between "concepts of things" and the
"things themselves". Where this really comes into light is when you
think about the assertions you would make a skos:Concept, do these
make sense when you apply them to the thing in question? For example,
when did Bob Dylan or the Grand Canyon become a "subject"? VIAF says
"2009-03-03T12:03:19+00:00" for Bob and id.loc.gov says "1986-02-11"
for the Grand Canyon.
Obviously the Grand Canyon wasn't "created" on Feb. 11th 1986 (and,
I'm sure, neither was the subject heading, but that's another issue),
but there's value in keeping the human construct of "concept"
separated from the object in question. The creation date, for
example, could tell a lot about when knowledge of a particular thing
came into being. dc:modified tells a lot about when perceptions were
changed :)
> Speaking of which, I took a look at FOAF and RDA/FRBR person entities, and
> the only overlap is that they both have names. I'll try to write it up, but
> it looks to me like we might be able to use FOAF for "social data" about our
> identified persons (mainly contemporary ones), and thus extend RDA, and RDA
> can help ground FOAF with more facts (date and place of birth, etc.).
> There's more... will work on it.
I definitely agree this can cover many schemas and different domains
will have different contributions on the matter. An environmental
scan like you're talking about here would be invaluable (don't equate
this with "priceless" :)) to find the ways to integrate ourselves
"with" the web rather than "around" it (which is what the RDA vocabs
tend to lean towards now).
-Ross.
Received on Tue Apr 13 2010 - 11:58:33 EDT