After testing some specific examples about deduping contacts with the deduper extension, I came to the follwing conclusion :
diatrics / accent / lower and upper case are very well taken into acount. For example, these 2 contcts are detected as duplicates :
First Name Last Name
François-André golD-sMith
Francois-Andre Gold-Smith
But the hyphen between gold an smith prevents duplicate detection in the following example :
First Name Last Name
François-André golD-sMith
Francois-Andre Gold Smith
As per this documentation page,
The Uninformative characters resolver. This strips a range of white space and punctuation characters out when comparing names. Currently the list is hard coded but I'm open to making it configurable. It also has a shorter list of characters that it will strip only if that resolves the conflict. For example a '.' is stripped in the uninformative characters resolver as that will mean later the initial resolver has a better chance of working.
From my point of view, in the above example, the hyphen and space in the last name should be treated as uninformative, so that they are detected as duplicate. Does it make sense ? Eileen, is that something you could add in the list of hard coded characters ? I would also add dash, underscore and tabulation at the same time if not already implemented...
I am using a painfull workaround : use the "external ID" field where I concatenate LastName and First name in lower case without diatrics nor uninformative characters "goldsmithfrancoisandre" I have to export to excel, concatenate in excel and import back to civicrm. This is painful but it prevents duplicate
Not sure if it's the right thing to do, but I have also posted the same here : https://lab.civicrm.org/extensions/deduper/-/issues/1
Thx!
