> -----Original Message-----
> From: tang2_at_simmons.edu [mailto:tang2_at_simmons.edu]
> Sent: Thursday, July 19, 2007 4:05 PM
> To: Next generation catalogs for libraries; Andrew Nagy
> Cc: NGC4LIB_at_listserv.nd.edu
> Subject: Re: [NGC4LIB] Open Source OPAC - VUFind Beta Released
>
> This is indeed a fascinating and interesting OPAC. After some initial
> exploration, one thing that I found to be a little problematic is the
> "Similar Items" Function.
>
> What was the similarity algorithm based on?
>
The similarity seems to be somewhat problematic as many people have already found.
Right now the query is done as a normal Solr query and does not use the 'More Like This' patch that is available.
The query matches on the following fields based on the record that you are viewing:
Title^3
Callnumber^10
Topic^5
Language
Author
PublishDate
The ^ character defines a value boost.
I played with the query a bit to get it as accurate as I could, but this still leaves many problems and should be re-evaluated. But hey, it's open source. So please feel free to edit it and make it better!
Any suggestions on increasing the accuracy would be appreciated.
Thanks
Andrew
Received on Thu Jul 19 2007 - 14:18:47 EDT