[AccessD] Fuzzy Matching (Like Soundex) or other ideas?

[Rocky Smolin]

Would stripping all the special characters out and matching just letters
and numbers get you there? You'd probably need another field in the table
where MW-14 or MW 14 appear, to hold the stripped down version of the data
so the search would be fast.

r

On Fri, Apr 21, 2023 at 5:27 PM Ryan W <wrwehler at gmail.com> wrote:

> Does anyone have any fuzzy matching routines that would have fuzzy matching
> logic based on a matching percentage or something else?
>
> Example:
>
> Client sends us a hand written chain of custody, they list something we
> look at as:
> "MW-14" so we enter it as MW-14
>
> On a set of forms they provide later for us to use reporting purposes, they
> called it MW14.
>
> So now we have data that doesn't precisely match.  The usual fix is for us
> to ask the client which one is right, and fix the incorrect one.  Sometimes
> it's the hand written form (and then our database, because we relied on the
> hand written form to start the work), sometimes it's the data entry forms
> they provided (You figure they'd get this right....).
>
> I'm trying to make it so when they send us the data to digest toward the
> end of the job, I can pull that data in and it'll match MW-14 with MW14, or
> vice versa.
>
> While SoundEx works for MW-14 vs MW14, it also thinks.. .. as an example
> that MW-104 matches as well. (or something similar as an example).
>
> I'm not even sure if a percentage match would be enough since MW-14, MW14
> and MW-104 are all a really tight grouping of "like" characters.  MW-104
> would be unrelated to MW-14 or MW14... so erroneously matching it would
> cause more grief than us just hand patching the IDs before we button up the
> job.
> --
> AccessD mailing list
> AccessD at databaseadvisors.com
> https://databaseadvisors.com/mailman/listinfo/accessd
> Website: http://www.databaseadvisors.com
>