in reply to Merge/Purge address data

I think the best way is to use something that works for each part specifically.. (Eg something for a name, surname, address, zip etc) and which you can teach which parts should be regarded as equivalent.. (eg MD = Dr., A. can equal Alan, etc.)

As someone else said, comparing bits which arent likely to change is a good start, the house number, the surname, the zip code..

Good luck with this, BTW, my company sells a pretty expensive product to merge telephone subscriber data doing this sort of thing.. :) - Maybe I can find out our approach.. (I don't work in that dept.. but.. :)

C.

Replies are listed 'Best First'.
Re: Re: Merge/Purge address data
by cleverett (Friar) on Nov 11, 2003 at 11:55 UTC
    so:

    1. parse names and address into more discrete chunks
    2. consider some columns more relevant than others

    Boy, 2 looks like a fuzzy logic thing to me ... maybe some kind of scoring system ...