Hi guys, I need to do name matching.

When there are differences at the end of the name, e.g., Martin/Martino, I can find an identical match by reducing the sensitivity

huks_0-1756295306745.png

huks_1-1756295387957.png

but the problem is when there are differences at the beginning, e.g., Fikir/Sikir.

huks_2-1756295562092.png

huks_3-1756295656195.png

How should I configure it to find an identical matchcode, because the sensitivity always reduces from the end to the beginning? There is just 1 letter difference between the two and I can't find the match!

Thanks for your help

Hugues

Maybe a combination of an address match and surname spelling difference might be a simpler option than tinkering with the QKB in that case. I've combined techniques like this myself and they can be quite useful.

I'm pretty sure that letter differences at the beginning of a name are much more important than at the end. For example Bell and Dell, Ball and and Wall are clearly different surnames so shouldn't have the same match codes.

Maybe a combination of an address match and surname spelling difference might be a simpler option than tinkering with the QKB in that case. I've combined techniques like this myself and they can be quite useful.

yes i already do that with firstname and lastname but i think i'll change the sensibility level thanks

You might want to customized the QKB definition to fit your needs if the existing one doesn't.

Take a look at this video for a quick introduction
https://www.youtube.com/watch?v=TdvKDVdBrdk

Hope this helps,

Audrey

If your goal is to identify strings that have just 1 letter difference, maybe it's easier to use a distance function in an Expression node.

Because when you make changes to the matching definition, it can have a large impact. For your example, you could be updating the Phonetics library to say "F" ans "S" are similar in sound, and that would fix this specific issue. But it would have a much bigger impact and could trigger lots of false positives.

I don't think there is a easy fix without knowing your data and requirements.

Audrey

thanks for your answer. I can do that indeed but but what I want to try is to make a matchcode reading from the left to the right and not as usual, from the right to the left . Because as i explained before, if I have Martin and Martina or Allan and Allen . i can find a near match playing in the sensitivity but with the difference on the first char i can't.

Like @SASKiwi already mentioned a difference in the first letter is normally relevant and the match codes need to be different. If you have cases where they should be the same then you need to learn how to amend/extend the out of the box QKB to cover your special cases. This should become a BAU task if using SAS DQ in earnest.