Match Similar Strings - Fuzzy match?

[KenReay]

Hi

a general query does anyone have any experience of matching (for example similar names and addresses), so that (say) 1 King Street and 1 King St are recognised as the same thing,
or even close misspellings, say Whitchurch and WhiteChurch, and recongnised as probably being the same.

Any tips, useful links etc for doing this with Access and/or VBA? Regards

Ken Reay
Freelance Solutions Developer
Boldon Information Systems Ltd
Website needs upgrading, but for now -

http://www.kenneth.reay.btinternet.co.uk

UK

 

[MichaelRed]

there is a function "soundex", which exists in several versions and includes some language dependencies which works well for the fuzzy matching -at least to a point- While you can find NUMEROUS threads discussing it, thread701-326595 includes a complete function which I transposed from a U.S. Gov. (FORTRAN?) version. It works on a limited basis as the length of the input string is critical to the 'accruacy' of the output, so strings which are to short or to long tend to not be as unique as those with the 'expected' length. This 'means' that "St" and "Street" won't be very closely matched, but "Street" and "Streat" WILL be matched. The overall process depends on some simple assumptions, one of which is the use of "dipthongs" - which vary considerably between languages and somewhat even between dialects. Dipthongs are just pairs of letters which are used to represent a 'single' sound. Since I am quite a bit NOT a language expert, and you are (aparently) in hte UK, you should review this part of the code to assure yourself that it is appropiate to the situation.


MichaelRed
m.red@att.net

Searching for employment in all the wrong places

 

[KenReay]

Hi

Thank you for that Michael, I am as you say in the UK, so I will extract your code and experiment with same.

Regards

Ken Reay
Freelance Solutions Developer
Boldon Information Systems Ltd
Website needs upgrading, but for now -

http://www.kenneth.reay.btinternet.co.uk

UK