Another Metadata problem. From what I can see in the forum, this might take time
This metadata error is having an impact on Ancestry.com searches and has lead to some fairly confused associations being made. Here it is.
I have encountered an error in several US Census records for Staten Island, a borough of the City of New York. Staten Island is variously known as Richmond County or the Borough of Richmond. There is also the town/neighborhood of Richmond in Staten Island.
The error: The census for 1900, 1920 and at least one other have placed people living in the Borough of Richmond (clearly written at the top of the form) as living in Richmond, Ontario, New York, which is, as the name implies, in northwestern New York state near the border with Ontario, Canada some 330 miles away. As I mentioned earlier area is affecting searches within other systems, such as Ancestry.com. I have been able to change some of the records for individual family members I am researching but I feel there needs to be some what to make a suggested correction or edit to the entire document directly.
Please see screenshots of random entry from 1920 with highlighted error, as well as a screenshot of entry with corrected field.
Can this be corrected in the 1900 and 1920 census and the error searched for in other census records for Staten Island? As I mentioned before, it's adversely affecting search in other databases, too. If not, can family search add a feature which allows users to at least openly suggest corrections publically within documents or offer a system for publically flagging suspect errors?
Answers
-
The Richmond (Staten Island) vs Richmond, Ontario problem has been reported and documented several times.
1 -
I'll say it again - could it possibly be that the use of AI and/or "standardizing" place names is part of the problem??
0 -
Thank you Aine. So it's being work on, I guess.
I hoping that by bringing up this particular problem again, more energy will go into correcting it. Do you know if there is a way I can volunteer for this particular project?
On another matter, apparently my images were in violation of community standards. I have read the standards and I don't understand what I did wrong. Think they will be take down but since they are aware of the issue I guess illustration is not needed. It just upsets me that I might have upset anyone somehow.
0 -
The repair is not being crowd-sourced. The Engineers/Developers are working on a system-wide fix. It's not just 2 Richmonds in New York. It's most places that have same names. See, for example
1 -
@H T14 You are interacting, initially anyway, with a computer 'assessment' of your image. When you receive an email highlighting the issue, there should be an option for you to get a human opinion
1 -
Mark, sounds like you are referring to a rejected Memories issue. That's not what we're discussing here. HT14's images were not blocked.
2 -
Aine I was told in a direct message that my images were rejected but I see they are still here. I was just concerned because I'm new here and want to get off on the right foot. Thank yo Aine an Mark
1 -
Every image is reviewed (by a bot) before it can be seen. Sometimes the system is slow.
The option to request human review applies to Memories, not the Community.
All is good.0 -
@maryellenstevensbarnes1 asked
"... could it possibly be that the use of AI and/or "standardizing" place names is part of the problem?? ..."
AI has nothing to do with it because the issues were cropping up long before AI was being used by FS (or perhaps I should say, long before they told us it was being used)
The issues are, so far as I know, part of the standardising process - which Richmond is this Richmond? You don't need AI to do that, you just need an old fashioned software algorithm. (I have deliberately phrased questions on here to give people the opportunity to admit to the use of AI and so far no-one admits to it. But as I say, an ordinary algorithm will suffice for place and date standardising)
What the issue appears to be is that the algorithm needs to cater for all sorts of real world names and any changes need to improve things without fouling up others. No point in fixing Richmond, Staten Island if it results in Richmond, Virginia being standardised as Richmond, Yorkshire.
I suspect that the algorithm is now layer upon layer upon layer. I have known software like that. I tended to eventually lose patience with it, rip it up and start again but it's not my software now, is it? 😉 In any case, a tiny, tiny error rate (and I never wrote an error free program) will still hit a large number of placenames, which Mr Murphy will ensure are my placenames! Frustrating though it might be, we need to be patient.
2
