Welcome to CSU Systems Blog Systems Group Blog

Add an New Entry |  Site Search |  Wiki

Live or Quick Searches
Posted by Edward Iglesias on Wednesday December 21, @09:12AM from the dept.
An interesting post on outgoing. Really fast predictive search of OCLC.

Work in progress

I've been talking about 'live' or 'quick' search for a while now.  I finally got a demonstration (or maybe here if you have trouble accessing non-standard HTTP ports.  There's another file you might recognize here).  I'll describe that a bit, but really more interesting is what we might do next.

What you should be able to see is an index to all the records that a large public library holds in WorldCat.  We've extracted all the 5-word phrases from authors, titles, statement of responsibilty, and subject fields.  It's a bit of trick to get the right phrase from the right manifestation from the right work to display.  We get the speed by loading all the information into memory in several flat files, and generating the screens from those.

The screen shot above, though, shows a prototype we're working on that's quite different.  For one thing, most of the information is in a Pears database, although we've done a lot of precoordination to reduce retrieval time.  That's working well, even though there's a lot more going on with SRU replies in XML being converted to HTML with XSLT.  More important though, is the categorization we're trying to display on the left.  A couple of weeks ago Lorcan suggested to Diane and I that we should try to combine the features of the Dewey Browser and Live Search.  So, we've been thinking about it, and this is what we've come with so far.

As we decide what citations to display, we compute the most popular DDC categories associated with all the records that match the search.  The plan is that the user can then interact with the subject categorization much like the Dewey Browser works, except that the most popular records will display as you click on the captions.

We'll see if we can make all this work.  In addition to unaddressed interface issues, there are some real problems scaling this to 60+ million records while keeping the interaction 'live'.  Jenny is working on this, though.  Thanks, too to Ralph for database support.



TrackBack URL for this entry:

Listed below are links to weblogs that reference Work in progress:

» Cool search from Lorcan Dempsey's weblog
Thom has a note about some really nice stuff on his blog. This is just a note: check there for the details of how these work. Look here for a 'quick' search of the Phoenix Public Library collection. I really like the way this works - it suggests answer... [Read More]


Open Archives Best Practices | Innovative's New WebPAC Pros  >


Related Links
  • Articles on Web Development
  • Also by Edward Iglesias
  • Contact author
  • ( Reply )
      Squishdot Powered   All my life, as down an abyss without a bottom. I have been pouring van loads of information into that vacancy of oblivion I call my mind. Logan Pearsall Smith (1865 - 1946) US-English essayist, editor, anthologist