Source linker messes up standardization of places
I found you have to check carefully the standard place source linker offers after adding it, as it sometimes scrambles or omits part of the place. At times, a completely different place is shown. Workaround is to copy and paste the source place on the left into the standardized place on the right.
Kommentare
-
This is a known issue. We ask you to be patient while it is reviewed and corrected. You should look at the underlying image if available, as there may be additional details that were not indexed that could apply.
1 -
Yeah, always check the actual document if at all possible, because what's showing up in the index may have no relation to what the record says about "where". (And what the index currently says may have no relation to what was originally entered for the location; I have a theory that some of the mangling that we're attributing to Source Linker is actually a lack of communication between it and either the automangler or the index correction disfunction.)
2 -
It's having a serious problem tonight.
The record I'm working on is from Brooklyn, New York City. The source linker is creating a place:
New York City, New York City, Brooklyn, Assembly District 16, Brooklyn, 24-1754, Kings, New York, United States1 -
@Aine Donnghaile
Thank you for reporting it here. I will discuss it with the engineers and they will advise me when progress is made.1 -
I gave up trying to attach anything last night using the source linker. And today hasn't been much better.
0 -
Ok folks I need some help to understand what's going on here, so that I can change the way I work to get the right end result.
In many of the England and Wales census records the place names as presented are mangled. The worst case being 1851 in many parts of Yorkshire.
When I attach a census source to a person, I try to correct the place name to be as accurate as possible without allowing the auto-mangler to change it to a near miss.
Recently I've noticed that although the place in the field that I'm filling is correct(ish), and the standardised place name shown in the field below it is an acceptable approximation, after the source is attached, the placename written into the standardised residence field of the person record is not the one that was shown in the standardised field by the source linker.For some bizarre reason the standardised place has reverted to the auto-mangled placename that was wrong before the edit. I have to go into the Person details for each person in the family to change the residence data to re-correct the standardised place again.
This is new behaviour that has only recently started.
How can I make these corrections and avoid this strange reversionary behaviour ?
UPDATE: Possible connection to the fact that all of Southowram has now become Halifax St Ann in the Grove, Halifax, Yorkshire, England, United Kingdom in the 1861 census. I think there's been an A.I. at work going through the place names.
1 -
Follow-up to my original post: As some have noticed, 'auto-mangle' is sometimes 'auto-fix', in that, the standardizer goes back to some original source place entry and uses that instead. I have found that instead of mangled, it's more correct than what showed originally. I suspect this all goes back to what is displayed in the place field in various parts of the program, the standardized or original, or something along the way, you never really know. And, to how it was originally entered and saved. For example, a cemetery in Chicago IL USA may be 'St Agnes Catholic cemetery' or some such; this can show up in the source or its display as something like 'St Agnes of the Well, Berkhamshire, England UK' (just made that up). The moral is, check your place names carefully: standardized or not, you need to learn some geography to be sure they're at least sensible.
0 -
Following the recent 'improvements' in the community threads, I'm adding a comment to this thread to see where it appears in the list of comments.
Scott's right that the mangler sometimes fixes, but on balance it needs a lot more training. I'm guessing that the average age of contributors is more than fifty times the age of the A.I. that's trying to learn to fix place names. The first months of life are mostly food and sleep mixed with mystery, fun, heartache, and getting rid of waste products. The A.I. can skip some of those, but if it's not learning from it's mistakes, then it needs some special care, and maybe a different environment might help.
0
