I think about using FunctionQuery. But as I understand, the FunctionQuery actually overrides the default similarity function. Am I right? If I am, this means that I cannot access the similarity value, which is necessary, because I want it have a major influence on the final score. I eventually need to calculate the score using a formula having several arguments. One of them is the default Solr similarity, i.e. the relevancy of a document for the specific query. Other arguments are derived from the original document field values.
Thank you. -----Original Message----- From: Otis Gospodnetic [mailto:[EMAIL PROTECTED] Sent: Thursday, October 25, 2007 4:42 PM To: solr-user@lucene.apache.org Subject: Re: Score customization Victoria, Either use FunctionQuery's or hack around HitCollector.collect(int, float) in SolrIndexSearcher...and adjust the score using the additional values you mentioned. Otis -- Sematext -- http://sematext.com/ -- Lucene - Solr - Nutch ----- Original Message ---- From: Victoria Kaganski <[EMAIL PROTECTED]> To: solr-user@lucene.apache.org Sent: Thursday, October 25, 2007 7:19:46 AM Subject: Score customization Hi! My system uses solr to build a searchable archive of documents. I need to override the default scoring/similarity function because the additional to the query relevancy factors have to be considered. For example, each document has "updated on" and "source" fields, which should influence the score too. I want the score/similarity function be a sum of weighted (according to configuration) query relevancy, update level and source rank (calculated at search time). Which is the best way to achieve required functionality? Should I inherit and override existing code or should I build some independent module (something like requestHandler) and to configure Solr to use it? If I have to override the defaults, which class should it be? In general, what am I actually trying to customize: scoring or similarity? Thank you very much. Victoria
