Okay, I only bought your book a few days ago, so I haven't read much yet! Also, this project is still at the exploratory level, so very little time has been dedicated to it (the irony is I am currently spending most of my time trying to work around K2 ... issues). As such, it will be some time before I get my hands dirty with actual code (my finger tips are itching, however).
There are further complexities, also. For example, if a query includes a field search, then it will be important not to match the name of the field in a given document. I _don't_ think the solution will come by simply indexing the queries as if they were short documents, but how the commercial guys do it is (of course) a black box. Thank you nonetheless for your ideas, -- Robert On Wed, 16 Mar 2005, Otis Gospodnetic wrote: > I think this is doable. Lucene compares how similar query terms are to > terms in indexed documents. Flip that around, and you are close to a > solution. Perhaps you'll want to process incoming documents to keep > only the top N most-important terms or phrases. You could then create > a query out of those terms and run it against indexed queries. > There'll be some tricks, such as dealing with situations where your > query contains a phrase, and you have to prevent individual terms in > incoming documents to matching individual words of that phrase... > > Tell us how this goes, I'm interested in this, too. > > Otis > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
