Well I was asking it because I have a custom FieldComparatorSource that uses lucene score among other params to calculate the sorting. The thing is that with my own lucene servlet I am getting different results than using solr now (because score values are different and Solr is giving me back the same score for somre docs).
Erick Erickson wrote: > > Why do you care? I'm not being too much of a jerk here, becausescores > between separate queries are irrelevant. See: > http://wiki.apache.org/lucene-java/ScoresAsPercentages > > <http://wiki.apache.org/lucene-java/ScoresAsPercentages>So, the scores > aren't important, the important thing is whether > the documents are returned in the same order.... > > I realize that you think you're doing the same query, but are > you really? Have you captured the actual *Lucene* query that > comes out of dismaxrequesthandler and tried using that one > exactly as it is used in SOLR? > > HTH > Erick > > On Fri, Jul 10, 2009 at 10:35 AM, Marc Sturlese > <marc.sturl...@gmail.com>wrote: > >> >> I have noticed a weird behabiour doing score testing. I do a search using >> dismax request handler with no extra boosting in a index of a milion docs >> searching in five fields. >> Printing the score of the docs 3th,4th,5fh,6th I can see that is the >> same. >> If I build the index with my own lucene indexer and print the scores with >> my >> own lucene servlet (building the same query with boolean queries)... I >> can >> see that the scores are diferent. There are no docs with exactly the same >> score. >> Why there are in Solr? Is the score suffering any modification before >> being >> added to the XML?? >> >> -- >> View this message in context: >> http://www.nabble.com/printing-scores-tp24428337p24428337.html >> Sent from the Solr - User mailing list archive at Nabble.com. >> >> > > -- View this message in context: http://www.nabble.com/printing-scores-tp24428337p24429699.html Sent from the Solr - User mailing list archive at Nabble.com.
