Can automatic location matching be more location-aware?
Hi. I just came across a situation where the source linker matched a location on a record to the wrong global instance of the locality. It's not the first time. In fact, since a great many jurisdictions don't bother to qualify where a municipality is unless it's "somewhere else" (as in this case), it's a big, big problem.
This problem has become even more time consuming to deal with because the Web UI no longer provides an opportunity to review and correct these mistakes. I need to go back later to the profile to do that. When it's been repeated over many records in a census, it's a big task.
Below is a screen shot. As you can see the death took place in Pembroke and, since it's an Ontario, Canada record, it's clear this is Pembroke, Ontario, a small city of about 14,000 people. Even though this Pembroke and several dozen others pop up in the locations database, it instead matched it to a somewhat different name, Pembrokeshire. Why would it do that? Is that locality more "popular", perhaps? Clearly there's something more than simple name matching going on here.
Please, if there's no state/province/country to go by, assume localities are in the same jurisdiction, rather than doing a global search. It's the sensible thing to do.
Answers
-
Autostandardization strikes out again.
For some reason, FS ran this bot with apparently zero data validation: the process just picked whatever came up at the top of the list, with no reference to the collection's metadata about where those places actually were. This means that you can't believe anything that FS says about "where": the only reason they're always on the correct planet is that places on the Moon and Mars aren't in the database.
If you could provide a link to a specific instance of the error, then a mod can add it to the list. I don't expect this to have any noticeable effect within this decade, but it's what FS apparently still wants us to do: report the errors so they can fix them piecemeal.
1 -
I included a screenshot to show the problem though for some reason the community web site isn't able to access it now. This specific page is here https://www.familysearch.org/ark:/61903/1:1:JND5-S4D . Going by what I can see, the Birth Location is just as on the record, "Pembroke". If there's some "actual" location stored with it, it's not visible.
Oddly enough, the record appears to have been corrected by my earlier attachment. I detached it and removed Ivan's death information, then went through the attachment dialog again and it now says Pembroke, Renfrew, Ontario. Seems to learn.
0