Find similar names

[Bekah]

I've been playing with ideas to find customer names that are similar. This is because we have many duplicates. Pattern Count doesn't work, ! doesn't work. I need something that finds similar ones that someone can then find to manually compare.

I've been playing with using Filter. It seems to me that if two fields are compared and contain the same letters (most of them) that they might be the same. I attached my play file. Am I on the right track here? Can anyone see what my thinking is, and would something like this work?

Or maybe I should include their address also, although someone could have moved. Any ideas to help me find similar names would be wonderful. Also, this file won't work because the fields are in the same record but they wouldn't be ... they would be different records. Big bad bummer.

Well, I think that if I remove all letters that match, and only a small number of letters don't match they might be similar. What do you think? Ideas on attacking this problem would be very much loved. Thank you.

test.zip

[Søren Dyhr]

We had several approaches up here: http://www.fmforums.com/threads/showflat...true#Post136377

--sd

[comment]

Soundex?

http://www.searchforancestors.com/soundex.html

[Ugo DI LUCA]

Hi,

I've been playing a lot with this concept too. Actually, there's a file, in fm5 format in the Sample section, called SpellChecker, which tries to catch mismatched names.

I'm working on a FM7 version now, involving a script which parse the content of the Left key.

Entering "Marlboro" will give a Basekey of

(|)(+)Marlboro

(|)M(+)arlboro

(|)Ma(+)rlboro

(|)Mar(+)lboro

(|)Marl(+)boro

(|)Marlb(+)oro

(|)Marlbo(+)ro

(|)Marlbor(+)o

(|)Marlboro(+)

M(|)arlboro(+)

Ma(|)rlboro(+)

Mar(|)lboro(+)

Marl(|)boro(+)

Marlb(|)oro(+)

Marlbo(|)ro(+)

Marlbor(|)o(+)

Marlboro(|)(+)

M(|)(+)arlboro

M(|)a(+)rlboro

M(|)ar(+)lboro

M(|)arl(+)boro

M(|)arlb(+)oro

M(|)arlbo(+)ro

M(|)arlbor(+)o

M(|)arlboro(+)

....

Marlbor(|)(+)o

Marlbor(|)o(+)

Marlboro(|)(+)

...

M(|)(+)rlboro

M(|)a(+)lboro

M(|)ar(+)boro

M(|)arl(+)oro

M(|)arlb(+)ro

M(|)arlbo(+)o

M(|)arlbor(+)

...

Well, a lot more as a base.

Then a sub-loop set a new key substituting each of (|) and (+) by one of the alphabet letter.

SetField[newKey;newKey&"

[comment]

Marlborough

[-Queue-]

It depends whether or not one is smoking.

[Lee Smith]

In Ugo's case, I think it is more of a What than if....

[-Queue-]

Especially after reviewing his newest sample file...