[Web4lib] Re: Google Search Appliance and OPACs

arhyno at uwindsor.ca arhyno at uwindsor.ca
Tue Feb 12 11:12:20 EST 2008


I have never used the Google Search Appliance but Google Desktop is 
capable of indexing a library's collection of metadata and injecting 
results into "regular" google searching. This approach is greatly hampered 
by the need to install Google Desktop in the first place, as well as a 
plug-in. Plus it can take about 8 hours of elapsed time for the indexing 
to occur for a million titles in an HTML format and it takes a good amount 
of compression to achieve this level of throughput  [1]. 

Still, one of my children is in second year at my institution and prefers 
this approach to our OPAC because of sheer familiarity. A highly tweakable 
indexing solution like lucene can probably run circles around this 
approach in terms of depth and technology, but the familiarity aspect is 
also hard to ignore. The combination that might have the most potential 
based on a familiarity quotient is using sitemaps and a google co-op setup 
[2]. I think this would allow for a "see the library catalogue in google" 
type of link on the library's web page, for example.

That being said, dumping metadata records into google can make for a 
really weak searching experience. As others have suggested, the key would 
be how to leverage linking in such an environment. For example, linking 
subjects and anonymizing patron records in order to create pages of 
borrowed materials expressed as links to indicate popularity. Our existing 
systems are potentially valuable repositories of such link information.

There have also been threads on this list about the limitations in 
utilizing Google Book Search [3]. Does a sitemap/co-op arrangement somehow 
allow a library to take advantage of the full text of materials that 
google has processed somehow? If the HTML representation of a title from 
the catalogue is somehow connected to the Google Book link, how much 
internal plumbing aligns for bringing them together in google's own 
indexing?

art
---
1. http://librarycog.uwindsor.ca/indexcat
2. http://www.google.com/coop/
3. http://lists.webjunction.org/wjlists/web4lib/2005-November/039133.html


More information about the Web4lib mailing list