When will the place name mangler get fixed ?
This has been causing problems for far too long now. It's become a real hindrance. The human effort involved in correcting the mistakes made by the machine is just not able to keep up with the errors. The original data is being disregarded and multiple successive changes are compounding to destroy the correct place data.
In the record I have just tried to attach the actual place is Smallthorne, Norton, Staffordshire, England, presented as
Event Place (Original) Norton, Staffordshire, England
with the Residence shown as Smallthorne
the auto standarisation has changed the Event Place from Norton to Little Norton . Little Norton is an entirely different place in a different part of the county. The key data is Smallthorne and yet the mangler has clearly ignored it.
Please stop messing with the data until there is a tried and tested routine that is capable of getting it right. The current one just doesn't do the job.
Answers
-
@SerraNola has taken the lead of compiling the issues and escalating to the appropriate team.
As I recall, we were asked originally to post what we noticed in the Search category.
0 -
@Re Searching Being that the "auto-standardizer" tried to match what was indexed with a standardized name, it was inherently flawed. Norton, Staffordshire, England is not a standardized name. Neither is Smallthorne, Norton, Staffordshire, England. Thus, what you get is Little Norton, Staffordshire, England. Of course, knowing why it happened and getting it fixed are two different things and it is more complicated than just this one type of place name error.
I give credit to the Community for helping us communicate to Engineers where the greatest problems are and the impact for researchers. It is especially important to report collections that have not already been discussed at length in various threads.
1 -
@SerraNola The reason why the auto standarsization routine selected Little Norton, is because in all the synonyms it has collected, L comes before N and S. That appears to be part of the problem. When there is no clear distinction between the various synonyms it should leave the data alone and not replace it with something vaguely similar that comes earlier alphabetically. If there are say 10 synonyms, then the chance of the first one being correct is 10%, that's just not good enough.
What's worse though, is that when it automatically changes the place into a different one while linking, it does not provide the drop down list of optional alternatives. So it switches Norton, Staffordshire into Little Norton, Staffordshire, and the user has to manually correct the change that should never have happened.
1
