You can find out exactly what is going on with Solr's debugging capabilities. Follow these steps:

1.) Turn on debug in your config.ini and repeat the search -- this will output the Solr retrieval URL for the results.

2.) Copy and paste the URL into another browser tab; you may need to change "localhost" to the name of your Solr server.

3.) Add &debugQuery=true to the end of the URL, and change wt=json to wt=xml for easier readability.

Now if you scroll down to the bottom of the Solr output, there should be a debug section showing exactly how the relevance scores were calculated. This may offer some clues. If you need help interpreting results, feel free to share the file.

- Demian

From: Gregory M. Testa [gmtesta@infopeake.org]
Sent: Friday, April 25, 2014 10:50 AM
To: vufind-general@lists.sourceforge.net
Subject: [VuFind-General] Question on Search Results

Just a quick question on search results relevance rankings.  If you do a search of The Help in my catalog instance (http://ezlibrary.infopeake.org ) either with All Fields or by Title

you will get hits on the two items whose title  consist solely of the short title, but then I get some other results after that with Help in the short title and “the” in the 245h info, and then it will list other items for The Help – CD sound recording, DVD recording etc. lower in the rankings.

This seems odd to me as I have the title_short field with the most weight at the top of both All Fields and Title in the searchspecs.yaml file.  I went and removed “the” from the stopwords.txt file, but that didn’t make a difference.  It seems like all my The Help entries should come to the top based on title_short (it lists them that way with the autosuggest feature in the searchbox).

 

Thanks,

Gregory Testa

Systems Analyst II/Applications Lead

Chesapeake Public Library

Information Technology Department