<<<is that the terms have to be in order to be considered a hit>>>
This is not true. "slop" includes re-arranging the terms, it just takes a little more slop (see Lucene In Action for an excellent pictorial explanation). Best Erick On Mon, Jun 13, 2011 at 10:45 PM, mtraynham <mtrayn...@digitalsmiths.com> wrote: > Hey Erick, > > Thanks for the feedback, but I it doesn't particularly solve my problem. > The issue with doing a boosted phrase clause is that the terms have to be in > order to be considered a hit. I'm seeking a solution where the terms can be > near another term in any direction. > > If I were to use a phrase clause, I would have to permutate every possible > ordering to get all possible solutions. i.e. Tom Cruise Dancing, Tom > Dancing Cruise, Dancing Cruise Tom, etc. > > I did try using a TokenizedPhraseQueryNode, which takes a phrase and breaks > down each word into FieldableNodes. I did have luck with passing most of > the later processor mutations, but some still affected it and therefore made > parsing each field node into a SpanQuery pretty hard. > > If I can make a FieldableTokenizedPhraseQueryNode to have untouched > FieldableNode children, then I could translate all the children into > SpanQueries and put them into a subsequent NearSpanQuery at the builder > stage, but this is still pretty incompatible with most of the pipeline. > > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/AndQueryNode-to-NearSpanQuery-tp3061286p3061607.html > Sent from the Solr - User mailing list archive at Nabble.com. >
