360 Search — it’s not bad!

Just back from a demo of “360 Search”, a Federated Search product from Serials Solutions. Now while I have previously said that Federated Search is a load of dingos kidneys, and will never work, I have to say I was quite impressed with this offering. It still suffers from the same problem that all FS products share, namely the dependence on external providers’ response times affecting the local response. But the demo searched over 20 databases simultaneously and brought back a consolidated 450+ result set in around 60 seconds, so that’s not bad performance, considering.

One very nice feature here is the inclusion of results clustering (using software from Vivisimo), which is great, and makes narrowing of the result set easy.

We were looking at it for a very specific and narrow purpose, but … despite my previous experience and subsequent loathing of FS … I think this is good enough to put in front of students.

Time will tell.

Blogged with Flock

Library catalogue enhancements revisited

I’ve revisited the code I developed last year, which adds useful links to our library catalogue records. It was always somewhat experimental (perpetual beta) so the revisit is not surprising: as I learn more and understand more about what people want, I’ll no doubt revisit again.

This time round — prompted by my Facebook investigations — I’ve added a few “share this page” links, to del.icio.us, Digg and Facebook.I’ve also revamped the very ugly code that provided a bookmark-able link for an item, turning it into an “Add to Favorites” button. This uses a nifty bit of Javascript found on the web, which works both in IE7 and Firefox 2.0-based browsers.

We still have the previous enhancements — Amazon covers, links to Google etc. Although the code has undergone a major makeover, basically a tidy-up to make it public-ready.

You can see an example catalogue record here:
http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=642088

And the code is here:
http://library.adelaide.edu.au/local/scripts/andjagetta.pl

Blogged with Flock

DIY Mozilla/Firefox/Flock search plugin

I just spent a jolly hour creating a Firefox search plugin for our library catalogue. (Actually for Flock, which I’m now using — and lovin’ — for its integration with services like Flickr and Facebook. But I digress.)

Creating the search plugin was easy in most respects. There was plenty of documentation on the plugin format on Google. Unfortunately, all of it seemed to be out of date for Firefox 2.0. But the easy solution to that is to open the searchplugins folder in Firefox (or Flock) and copy one of the files there. Google seems a good one to use.Then edit the details and — viola — you have a search plugin.

Except: one of the requirements is an image file for the toolbar, and it seems that this image needs to be embedded in the xml file and Base64 encoded. I have an icon — the favicon used in our web site, but needed to convert that to Base64.

Now, there is at least one tool available on the web for converting files to Base64, but if you want to do it yourself, and can run perl, here’s a trivial bit of code that will do the conversion for you:

$ perl -we ‘use MIME::Base64; local $/; open ICO, “<favicon.ico” or die “$!\n”; $x=<ICO>; close ICO; print MIME::Base64::encode($x);’ > icon.txt

Then you can simply replace the Google icon Base64 code with the content of icon.txt, and you’re done. (Actually, you may have to remove all the line breaks.)

Blogged with Flock