I think that among other things we're struggling with some terminology
issues... "linked data" is a bit of a misnomer -- it doesn't mean that
the data *is* linked, it means it is in a format that could be used
for semantic web linking. More correctly it could be called "linkable
data", but then what would we call it when it actually linked?
Jim is really arguing my point, which is that the problem isn't with
what I guess we could call the data format, but with the content of
our data. LCSH at id.loc.gov is perfectly good, semantic-webbed,
URI-identified, linked-data compatible data. Is it being used as such
in the linked data cloud? No, not really. Why not? I think for the
reasons that Jim states: that the pre-coordinated forms just aren't
useful to anyone who isn't assigning standard LCSH headings. As Ross
seems to be pointing out, this seems to be proof that "just putting
our data out there" may not be good enough. It's a first step, we
should encourage LC to get the other authority data out there ASAP
(and I think names may turn out to be more useful generally than
LCSH), but we also need to know this is just a first step.
This is been a huge issue in the RDA registry project. You *can*
register RDA data elements in a semantic web-compatible format, but
they remain semantically library cataloging data. As I pointed out in
my talk at Code4Lib (2008?), the use of transcribed values, such as
the name of the publisher as it appears on the title page, is of great
interest to (some?) catalogers, but no other community appears to be
interested in capturing that particular bit of data. What other
communities probably would be interested in would be a linkable data
element for the corporate entity that is the publisher. The library
cataloging data does not include an element for that. Ditto, I think,
our compound headings, such as the uniform title heading that was
discussed earlier in this thread. Others may well want a linkable data
element with the original title. The uniform title heading, however,
doesn't give us this because it includes things like language and date
of the edition being described.
As a result of this conversation, which has been long, varied,
sometimes unfocused, but essentially very useful, I am more than ever
convinced that we have to do more than "link-ify" our current data
elements and data values. Note that the RDA registry project has taken
some steps to create data elements that can be used "in spite of" RDA.
We're working on an article to explain that, and I hope it will be out
by Midwinter.
kc
Quoting Ross Singer <rossfsinger_at_GMAIL.COM>:
>
> You're still missing the point. id.loc.gov has nothing to do with
> _pages_. It has to do with _data_. It's purpose isn't to add as a
> link to your webpage, it's to add a link to your 600 field so we're
> not doing weak text string queries across systems.
--
Karen Coyle
kcoyle@kcoyle.net http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet
Received on Thu Nov 12 2009 - 17:12:00 EST