Placename standardization errors in the new source linker
Attaching a Georgia Death Index to a cousin's profile LHNZ-F94 this morning, I saw this:
The record: https://www.familysearch.org/ark:/61903/1:1:V4M1-Z1N
I switched back to the old source linker to compare.
I believe the old source linker did it better. Automatically adding the absolutely wrong place (in new) is not an improvement over the warning message (in old).
項留言
-
Áine, thank you for helping bring this to our attention. It is a bug that we are seeing a lot of and we are currently in the process of fixing it. Thanks again for your feedback.
1 -
I dearly hope that the fix you are working on repairs this bug everywhere on FamilySearch where an incomplete place name is turned into a completely wrong place name such as in search results lists as seen here:
Where the incomplete Tysnes, Nor. which is most likely standing for Tysnes, Hamarøy, Nordland, Norway, was replaced with the totally incorrect Tysnes, Vestland, Norway. (This has a second error in that Vestland did not exist until 2019).
The on-the-fly auto-standardization routine responsible for what is seen in Áine's image and mine has a very poor track record for turning a single word place name into the correct full place name, particularly when the correct place name is not in the Places database yet.
1 -
After seeing Mike's comment about a fix in another thread (thanks @mikedavis2 ), I revisited the profile of John Henry Smith LHNZ-F94 to see what had changed.
I appreciate that the warning message is back. Thank you for that!
But, I'm going to stick my neck out and say that's not enough. I know this is the group for New Source Linker feedback, but the best Source Linker ever built won't help when we still have inherent issues from placename standardization.
This record is rather obviously a Georgia record. The options offered begin with PENNSYLVANIA. I don't know how that placename standardization algorithm was constructed, but there seems to be no discernible pattern. Fulton County, PA is not the oldest of the top 3; it is also not alphabetically first; and, it is surely not geographically closest.
Oh - and if I ignore the warning, it defaults to Pennsylvania. I detached and re-attached several times to test. I echo Gordon's wish
I dearly hope that the fix you are working on repairs this bug everywhere on FamilySearch where an incomplete place name is turned into a completely wrong place name
0 -
@Áine Ní Donnghaile @Gordon Collett
The fix for New Source Linker not saving Standardized or Edited Locations and Dates has been applied. Please help verify the change by reporting any observed anomalies.
0
