in reply to Re: Re: Merge/Purge address data
in thread Merge/Purge address data

Depending on how many addresses you have and if this is a one off you might be well off doing 4 yourself. You could write a script that would do 1, 2 and 3 and then interactively ask you whether to merge and what to merge.

In fact if you did this then you could add bits of code to do 4 as you went along. This way you will see the problems that are cropping up and will have a good idea of what is required to fix them.

Added: As I understand it baysian filters learn from experience. Maybe you could get the filter to look at you decisions above and learn from them. Potential to wander off into AI and expert systems here.

--tidiness is the memory loss of environmental mnemonics

Replies are listed 'Best First'.
Re: Re: Re: Re: Merge/Purge address data
by cleverett (Friar) on Nov 11, 2003 at 11:49 UTC

    I could cache the results of 5. When doing a match, the first likely match would lead to its known dupes.

    Added: As I understand it baysian filters learn from experience. Maybe you could get the filter to look at you decisions above and learn from them. Potential to wander off into AI and expert systems here.

    That wouldn't be so bad ...

Re: Re: Re: Re: Merge/Purge address data
by cleverett (Friar) on Nov 12, 2003 at 02:49 UTC
    EvdB said:
    Depending on how many addresses you have and if this is a one off you might be well off doing 4 yourself. You could write a script that would do 1, 2 and 3 and then interactively ask you whether to merge and what to merge.

    Didn't catch that at first. Actually, I'd want to run it daily ... no hourly ... make that as often as I can.