"optionally store document vectors in index"

[Dmitry Serebrennikov <dm...@earthlink.net>]

Doug,

Could you give me some pointers to the approach you think should be 
taken on this one? I'm in need of analyzing index terms from documents 
retrieved by a query to identify further query expansion terms. We 
already store the terms in a special field in the document, but that 
leads to very slow processing because documents must be retrieved, field 
re-parsed, etc. There are a couple of ideas that I'm going to try in the 
next couple of weeks to fix this, but if you have a roadmap for this 
already, I'd appreciate it. If the solution we end up with extends 
Lucene, I'll be happy to contribute it back.

Thanks.
Dmitry.