>> In point of fact if all libraries dumped their catalogs to linked data, some xml files, those files could be crawled by various services and in various ways they could link up the data to point someone to the library's holdings...
>
> The question is how this would actually be accomplished. Most
> libraries don't have this kind of expertise on hand, and even if they
> did, providing this service locally is more expensive than most people
> are willing to admit. That's why almost all libraries libraries still
> use a binary format developed for tape transfer in the 1960's to move
> their core data around. Anyone wishing to manipulate this data must
> use one of only a handful of tools -- the best of which were created
> by sole authors in their spare time. If everything is so easy, you'd
> think someone would just throw up a server. You'd also think MSU would
> just download records for free from public sources rather than getting
> them from SkyRiver.
As Tim Berners-Lee has been saying, "the first step of actually putting the data out there is the one that nobody else can do"[1]. We can discuss tools and methods and difficulties, but the point I'd make is that, right now, all of the data is locked up in libraries' ILL systems and thus totally inaccessible to anyone _except_ their own staff — so much the worse if they don't have the expertise (or time, or budget...) to do anything new or interesting with it.
How do we get there from here? Put your library's catalogue into the hands of those who *can* turn it into linked data and build services upon it. This is as easy as a raw MARC dump into the Internet Archive or other public repository — you don't even have to host it at your own institution. It really *is* that easy.
Yes, the data is dirty, yes control numbers are difficult, and yes the tools are sparse and incomplete. But by making the raw data available, you empower others to take up those challenges.
MJ
[1] http://www.readwriteweb.com/archives/interview_with_tim_berners-lee_part_1.php
Received on Tue Mar 09 2010 - 10:38:57 EST