New merge tool - good news and bad news
LegacyUser
✭✭✭✭
Paul said: I was heartened to see the "Merge Warning(s)" illustrated below. However, still somewhat disappointed that the FamilySearch algorithm is still causing two individuals to be offered as possible duplicates, apparently just because both had a spouse named Jane.
As can be seen, the one illustrated (of the two possible duplicates suggested) was not even born on the same side of the Atlantic!
Incidentally, anyone (probably me!) who intends to tidy-up this record (9CBY-PGQ) is going to face quite a challenge - careless users having added children covering a 48 year period and born at various locations. (Oh, I've just noticed one "child" is shown as born in Pennsylvania, so maybe that's where the possible US match comes from.)
As can be seen, the one illustrated (of the two possible duplicates suggested) was not even born on the same side of the Atlantic!
Incidentally, anyone (probably me!) who intends to tidy-up this record (9CBY-PGQ) is going to face quite a challenge - careless users having added children covering a 48 year period and born at various locations. (Oh, I've just noticed one "child" is shown as born in Pennsylvania, so maybe that's where the possible US match comes from.)
Tagged:
0
Comments
-
Tom Huber said: Yeah, but this isn't a problem with the merge tool, is it? Or am I misunderstanding what you've written.
It appears to be an on-going issue with the duplicate checker and problems associated with it. They certainly exist and it doesn't take much for them to follow the adageThe err is human, but to really mess things up takes a computer
That really applies when a routine attempts to use artificial intelligence to tighten its parameters.
True Artificial Intelligence is still a thing of Science Fiction. Despite claims to vast advances in the field, there are some uniquely human traits that have yet to be duplicated (and likely never will be).0 -
Paul said: No, not directly, Tom. But the two features are obviously interrelated and I was just illustrating that whilst on the "good" side we do now have a feature that gives an "extra" warning that the user might be about to make an incorrect merge, it is "another part of the system" that is suggesting the "bad" merge in the first place.
"Here, try this for a possible duplicate ("Possible Duplicate" suggestion on person page). Well, actually no - it looks like you're about to embark on a bad merge (Merge Tool message)."
So, you're partly right: only half my post (the "praise" bit) is DIRECTLY referring to the new feature!0 -
Tom Huber said: Incidentally, this goes beyond just the warning you illustrated (and praised).
I recently merged a duplicate but, the duplicate's birth record generated a warning message that more than three years existed between the duplicate birth record and the surviving record.
What was interesting (and the warning is a good thing) is that the surviving record actually contained a range for the birth year while the duplicate was precise. I believe the new system is going to stop a lot of the bad merges that previously took place, especially if place differences for vital information is also flagged.0
This discussion has been closed.