In article <[EMAIL PROTECTED]>, Paul D. Kraus wrote: > Ok this could work but then I would need to know before hand what words > are important and which ones arent. We carry over 40 thousand products > having to do with orthotics, prosthetics, ect. Unless I misunderstood > you. Thanks agian!
This sounds interesting (not that I think I could offer you a solution) ;-) but could you post just a couple tiny, itsy-bitsy samples of the data (even if it's been greeked a bit)? My own 'sample data'... (just to think aloud) Our description: "Firm, fruity apple, pale green color; with long storage life; excellent for baking." Their description: "Fleshy apple; is excellent for pies; long storage life; available until late November." Using your suggestion, you throw out: for, is, until In this case you would get matches on: apple, long, storage, life, excellent You could possibly improve this if you could prebuild a table of "equivalencies" (term?)... i.e. Keyword Table Keyword - possible equivalent matches BAKING - pie, cobbler, tart, bake NOV - fall, autumn, late Maybe this table is part of a self-learning program. Or, with the BLKKTL/Black Kettle mentioned earlier, maybe you can first detect a system of sorts (i.e. dropping vowels and double consonants)? Best would be to engage in some monopolistic price-fixing first and then you could match on price! ;-) -- Kevin Pfeiffer International University Bremen -- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
