lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Nina Khosravi <khosr...@us.ibm.com>
Subject CJKAnalyzer - Issues with scoring
Date Mon, 23 Jul 2007 04:38:20 GMT


Hello,

I am in desperate need of help on a problem with our newly deployed
application that is now using the CJKAnalyzer/CJKTokenizer.  Our previous
release was using the WhitespaceAnalyzer and the users were extremely happy
with the search results ordered by the score of the documents.  We switched
to the CJKAnalyzer and now the scoring is not giving them the results they
are expecting.   Any ideas on why using the CJKAnalyzer would be so
different?  Some documents that contain more instances of a search term are
appearing at a lower position in the search results than those having one
instance of the search term.

I tried printing the explanation of the scoring but don't get any
information, just the value 0.0 for each doc returned.  Anyone have any
ideas?  I have not dug deep into this problem.  I am pressed for time so
was hoping someone could provide some guidance.

Thanks!!

Nina

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message