How does a placename get attached to a Record Hint when it does not appear in the record?

Paul W

I came across this when the item was presented as a Record Hint. If you look at https://www.familysearch.org/tree/person/research-help/9KJ5-3XV you will see there appears to be an "Other" field within the record, mentioning the place "Plakje, Makedonien". (Second hint on Isabella's marriage.)

It's not just a case of how this place name gets attached to a record relating to a marriage in Yorkshire, England, but where it even comes from. It doesn't appear to relate to auto-standardization, as there are no place names in the record that could possibly be associated with it. I believe I have seen something similar in England & Wales census records, but no one appears able to suggest how such names / fields get attached to certain records.

Getting to the same item (exact same URL for the record) via a Search (rather than through the Record Hint)- see item 1 of 2 at https://www.familysearch.org/search/record/results?q.givenName=isabella&q.givenName.exact=on&q.marriageLikeDate.from=1851&q.marriageLikeDate.to=1851&q.marriageLikePlace=Hull%2C%20Yorkshire%20East%20Riding%2C%20England%2C%20United%20Kingdom&q.marriageLikePlace.exact=on&q.surname=bulmer&q.surname.exact=on) - you will see no reference to Plakje, Makedonien. (Unlike the examples I have encountered on the Results pages for E&W census records, which do mention a place not to be found on opening the record!)

In summary, is there any plausible explanation of how these placenames suddenly appear (in Record Hint items, Results items, or wherever), when the actual record / source - once opened - has no sign of it? (See - https://www.familysearch.org/ark:/61903/1:1:66GR-LXFX) Sorry, I keep getting a "Page Not Found" message when using this link, but the page can be found using "View Record", through either of the two links above.

Paul W

Success! I finally used the FamilySearch facility and the images (above) have been posted: a couple of hours after my request (using the form at https://forms.monday.com/forms/2f44be7ab074b1a5e1c3d6cfd9eb3074?r=use1)

I won't bother to repeat the exercise for this second example (below) - just provide the URL: https://www.familysearch.org/tree/person/research-help/GQR9-LND.

Looking at the third hint, you will see a mysterious piece of data: "Other: East, Westmorland, England, United Kingdom". Again, I cannot find this anywhere else apart from where the record appears as a Record Hint.

MandyShaw1

@Paul W I think it's a misinterpretation (no doubt typical of the current placename algorithms) of the residence place 'Vallace Place'. And, re your second example, the residence place is 'Roseland East'.

Julia Szent-Györgyi

(The last link in your post is coming up Not Found because the closing parenthesis has been "helpfully" included in it.)

I think MandyShaw is right that the on-the-fly automangler is putting Vallace Place in Macedonia, but I don't know why it's only showing in the otherwise-bare-bones Hints page summary. I also cannot fathom why the algorithm picked German for an English record.

I think the programmers need to go back to the basics: stop adding more and more levels of complication to the display, and simply show us a table of what's in the database. If it's in a list (like search results or hints), then naturally there needs to be some picking and choosing of which fields to display, but the chosen fields should simply be displayed, exactly as they were entered into the database. We do not need or want any of this automangling, and we don't enjoy trying to make sense of index detail pages where the same name can appear in like four or five different places, depending on who was named after whom.

Áine Ní Donnghaile

I also suspect it's the placename algorithm at work again. I recall one instance I reported, back when we could still see the "original" notation, that had turned the E.R. (for East River in NYC) into Eritrea.

MandyShaw1

I just do not see the point of all this placename mangling when the developers are very well aware that the algorithms simply do not work correctly. What harm could it possibly do to turn the placename interpretation off for all these 'display only' purposes?

Re @Julia Szent-Györgyi's user interface point, I completely agree. The 'other possible spouses' problems with E&W marriage records are another good example.

I imagine all this drives the support team up the wall.

It also worries me that these problems aren't being spotted at testing (or regression testing) time.

Paul W

Thank you all for your comments.

Yes, I could see the possible connection between the Rosedale East (in my second example) and East, Westmorland. But I can't possibly work out how Vallace Place could get to become any location in Macedonia!

However, the issue I am mainly raising here is how these placenames are only appearing on the Records Hints pages. Not, as with the example of England & Wales census records, where incorrect placenames did (also) appear on the Results pages (unlike in these examples), but - as with here - were not visible once the View Record link was activated.

BTW, Julia, I tried removing those parenthesis, but still got the "Page Not Found" message. Oh, and yes, whatever the reason it has been added, why not put the place name "Makedonien" in English?

MandyShaw1

If you put Vallace Place into the place box on Search Historical Records you don't get anything Macedonian, but if you put in Place Vallace you get 'Plakje, North Macedonia' as an 'option'. I can't get the Westmorland location to come up when I put in any combination of Roseland and East, though. It's a mystery.