Posting to Autocat
On Tue, 9 Nov 2010 10:31:20 -0500, Brenndorfer, Thomas wrote:
>Linked Data is quite different from regular HTML web linking or even from traditional database design. Linked Data is about sharing structured data using identifiers (which don’t even have to be HTML-based).
Yes, I know. I’ve done it. See below.
>> To institute linked data *right now* in our records, we could do it now and
>> don’t need FRBR or RDA.
>Encoding Linked Data URIs in MARC is riddled with problems, as indicated in this MARC Discussion Paper
There are LibraryThing APIs that can be used, Amazon.com the Internet Archive, HathiTrust, and so on and on and on. I haven’t worked much with these yet since I can only do so much here, but there are immense possibilities by using the information in our records *right now* and just being a little imaginative here and there.
One of the big problems in the library world is that there are no library APIs (that I know of) that are really useful for our public. It appears that it is possible to use id.loc.gov for linked data purposes, but you would have to add/change all of the current textual subjects to URIs, e.g. the heading “650 $aWorld Wide Web” would have to be changed in some fashion to http://id.loc.gov/authorities/sh95000541#concept and from here, you could apparently get the XML response that you could work with. I don’t know of any live implementations. It seems that there could be lots of possibilities here but things have to change around. For example, LC should be able to take the request, e.g. “http://id.loc.gov/authorities/World Wide Web” (easy to implement for almost anybody), and they would convert it to what they need: http://id.loc.gov/authorities/sh95000541#concept instead of expecting everybody in the world to change or edit their records locally. They won’t do that. If LC implemented something like this however, it would be much more understandable in a networked world, and could also be used very widely by non-librarians, and for non-library tasks, as well.
But one of the problems with this, even *if* this were implemented, is that the user would only see the authority record with, true, the BT, NT, RT, UF and scope notes, which are useful, but the only links from the record you see in id.loc.gov are to other internal links, so it is of very little use to the patron since there are no links into other items on the web. But in Bernhard Eversberg’s version: http://www.biblio.tu-bs.de/db/a30/lcsh.php?com=X%20a30ind–Vuf1=LCS–Vut1=world%20wide%20web (you have to click on “world wide web”) there are at least links into Google, Google Scholar, etc.
As an example of something that presents the user with a richer tool, look at http://dbpedia.org/page/World_Wide_Web from dbpedia. Don’t look at the formatting because that can be changed, you can display only the languages you want, etc. Look at all the different kinds of properties, e.g. “isInventionOf: Tim Berners Lee”. The “redirects” are the equivalent of UF. There are a lot of problems with this I agree, but there are lots of links to lots of resources out there. Which would an untrained user find more useful: id.loc.gov or dbpedia?
These are some of the promises (problems?) with linked data, but that is the world we should be getting into. And it can be done *right now* today. But we need imagination. Here’s a scenario:
Imagine a world where everybody follows ISBD–forget the punctuation, but the real meat of it: the rules for deciding what to input and how to do it. Then in the linked data “request” we could put in the ISBD information: the equivalent of 245abc, 250a, 260abc, 300a, 4xx, or whatever we wanted. The “reply” could be giving us the record for the manifestation we requested, or *anything else* we wanted, perhaps something entirely different, such as other subjects, or recommendations, or latest blog entries from academics, or who knows what. It could even supply us with the work/expression/manifestation FRBR displays if that were seen as so important. It could give it all to us.
Above all, the key to make this work is high-quality, predictable metadata that can supply a valid and predictable “request” and match it with a valid and predictable “reply.” Then the sky really is the limit. But first we must open our minds to absolutely new possiblities.