lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Donna L Gresh <gr...@us.ibm.com>
Subject MoreLikeThis and setBoost
Date Tue, 20 Nov 2007 15:35:23 GMT
I've been stepping through the contrib MoreLikeThis class and was 
wondering if people can give opinions on why you would or would not use 
setBoost(true) for the MoreLikeThis object. It seems a bit odd (at least 
to me) to boost the "good" terms in the query (based on the term's score), 
since won't the final score (once you use the query) in some sense 
"reflect" the effect of good terms already through the tf-idf? Is using 
boost in some way trying to "make up" for the fact that the returned query 
for the MLT object "loses" the term frequency of the terms in the 
reference document (that is, no matter how many times a term is in the 
reference document, the query remains the same, assuming that the term 
makes it into the query via the MLT heuristics).

Thanks for any words of wisdom--
Donna
Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message