Worldcat now on the web

OCLC have now put WorldCat (1.3 billion items in more than 10,000 libraries worldwide properly on the internet (rather than through Google or some other means). It says Beta although I don’t know if this is a proper beta, although the write-up and other documentation don’t seem to mention the fact at all, or a Google-style beta. Looks like the second.

The aim of the thing is clearly to enable one to find items locally. I tried to find Vromans’s Perl pocket reference somewhere near my postcode, and managed to get to a good set of results on the third screen: I didn’t even have to tell it that what I entered for location was a UK postcode (rather than a US zip code or a small town in Germany), a degree of intuition rare in library software. Cambridge University is actually not that far away really (28 km according to Worlcat), although the fact that the next two nearest libraries were Oxford and the British Library gives some idea of the paucity of coverage in the UK: the next nearest location is Institut informacijskih znanosti / Knjinica in Sloveniaafter which I’m looking at McGill in Canada and the Boston Public Library. All for so little a book.

Although not properly FRBRised, the interface is refreshingly helpful in narrowing down large results sets, having options that look a little like the gubbins on the left of this weblog. For instance, search for Hard Times and you get about 5300 results. The five most popular authors are listed on the left. Click on Dickens and you’re on a more manageable 156. Click on Book format and you’re down to 153. Click on English language and you’re down to 119. This leaves you with a further choice of dates. These choices are only the five most popular, which is a shame (well, it doesn’t clog the screen up which is definitely a good thing), especially in this case the top five dates are 1900, 1800, 1901, 1910, and 1880. A majority of all these hits are not actual editions from those dates, but unknown publication dates, where the cataloguer has put [19–?], or in one case [1—] (which means that the cataloguer knew which millenium the item was published but not the year, decade, or even century). For comparison, I tried our catalogue, which is for a large university research library (about 1 million records compared to 1.3 billion), and got 40 hits, had to go to a separate screen coyly marked Filter and use some pull-down boxes to get 6 results which looked on the nail.

For me, what seems very promising is what Lorcan Dempsey mentions: a simple syntax for linking. The example he gives is http://worldcat.org/isbn/0679454438 for The road to reality by Roger Penrose. If nothing else, this might be a better way of referring to a book rather than linking to an Amazon record. It looks neater, should be more accurate and complete, and even lets you buy from Amazon using a link on the right. It could be the book equivalent of the ubiquitous IMDB link. The only problem is I couldn’t see any documentation on how to do it on the WorldCat site. Incidentally, the aforementioned book can apparently be found in Bedfordshire Libraries, which gives the lie to my comments above about poor coverage, which applies only to Perl books.

WorldCat also lets one put a searchbox on your site, which is modelled below:

It should work, so do try it. All very good, although the scary list of terms and conditions and personal details needed for registration seems a reassuring return to a more-library-like fear of loss of control and openness.