Solr depends on Lucene's implementation of queries and how it returns document hits. I can't help you architect these changes.
On Mon, Jun 21, 2010 at 7:47 AM, sarfaraz masood <sarfarazmasood2...@yahoo.com> wrote: > Mr Lance > > Thanks > a lot for ur reply.. I am a novice a solr / lucene. but i have gone > thru the documentations of both.I have even implemented programs in > lucene for searching etc. > > My problem is to apply a new search technique other than the one used by solr. > > Step 1: My algorithm finds the tf idf values of all the terms in each url > and makes a chart like this : - > > term 1 term2 term3 ........... > url 1 0.7 0.6 0.7 > url 2 0.0 0.5 0.4 > url 3 > 0.7 0.8 0.6 > .. > . > . > . > . > (urls with 0 tf idf means word doesnt exist there.) > > This ways i first construct a complete chart of term tf idf to urls.. > > Step 2 (Searcher ) > then > depending on words in the query i select the correct urls by applying > mathematical formulae. This result should be shown to the user in > descending order. > > Now as i know that lucene has its own searcher > which is used by solr as well. cant i replace this searcher part in > SOLR by a java program that returns urls by my algorithm. Rest every > thing should be of solr. > > Only change the searcher part. I have > studied abt customizing the scoring which is absolutely not my aim.My > aim seems to be replacing the searcher. It is a work similar to BM25 > work which u had mentioned in your reply viz providing an alternate to > lucene search. > > Plz help me in this regards. I will be highly gratefull to you for your > assistance in this work of mine. > > If any part of this mail was not clear to you then plz lemme know, i will > expain that you. > > Regards > > -sarfaraz > > -- Lance Norskog goks...@gmail.com
