Mailing-List: contact solr-user-help@lucene.apache.org; run by ezmlm
Precedence: bulk
Reply-To: solr-user@lucene.apache.org
Received-SPF: pass (nike.apache.org: domain of lists@nabble.com designates
 216.139.236.158 as permitted sender)
Message-ID: <17367660.post@talk.nabble.com>
Date: Wed, 21 May 2008 11:35:28 -0700 (PDT)
From: peter360 <peter360@fastmail.us>
To: solr-user@lucene.apache.org
Subject: dismax handler and WordDelimiterFilterFactory
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit


Hi,

Let's say I have an index with two fields: f1 and f2, and queries to both
are analyzed using WhiteSpaceTokenizerFactory and
WordDelimiterFilterFactory.  I use dismax handler for queries and observed
the following anomally.

Suppose I have a document with f1="american" and f2="idol".  Then a search
"q=american+idol&qt=dismax&qf=f1+f2" matches.  However, the search
"q=american-idol&qt=dismax&qf=f1+f2" does not, even though the analyzer
(WordDelimiterFilterFactory) turns "american-idol" into "american idol".

Upon closer look, the dismax handler is parsing the first query as something
like 
+(f1:american f2:american) +(f1:idol f2:idol)
while parsing the second as something like
f1:"american idol" f2:"american idol"

I feel this is an anormaly because from end user point of view american-idol
should be treated the same as american idol.  How do I achieve this?  One
possible solution is to index f1 and f2 as one field, but I want to be able
to give separate boosts to them, such as "qf=f1^2+f2".  Any ideas?  Do
people feel this is a bug in the dismax handler?

-- 
View this message in context: http://www.nabble.com/dismax-handler-and-WordDelimiterFilterFactory-tp17367660p17367660.html
Sent from the Solr - User mailing list archive at Nabble.com.