In the Italian province of Udine most birth records display the birthdate as only a year
Full Question: How do I report to FamilySearch an issue where in an entire province in Italy most birth records display the birthdate as only a year, even when the full birthdate was indexed?
Quick Summary
Most birth records from 1875-1900 in the Italian province of Udine show only the birth year in the record, despite the fact that the full birthdate was indexed. This was initially functioning correctly, but sometime between October 2022 and April 2024 this problem appeared. Because of this, in Source Linker, when creating the child referred to in the birth record, only the birth year is added (unless the patron notices and goes through extra steps to find and apply the full date).. And now, it’s harder for machines and humans to disambiguate between two people with similar names who were born in the same year, contributing to the problem of profile and family “mangling”.
My estimate is that this affects upwards of 500,000 birth records in the province of Udine. A quick check shows that this is not a problem generally across all of Italy.
What is the appropriate way for me to report this type of widescale data issue in order to get it in the hands of the right FamilySearch team that can assess and fix the damage? (I suspect that the automated process that broke these will need to be reconfigured and rerun).
More Details + Screen Shots
In the “Italy, Udine, Civil Registration” collection (1939238), most of the birth records have a problem in that while the full birthdate was indexed (you can see this in the side panel of the image viewer),
only the birth year is displayed in all of the places where pieces and parts of the record are displayed.
Record Page
Search results
Profile sources tab
Profile Birth edit screen
Source Linker
For some reason, the full date is not getting passed along to the record.
This is especially harmful in Source Linker where new profiles get created with only a birth year instead of the full birth date. If a patron is careful and experienced as they are creating the child in Source Linker, they can navigate to the image, and copy the full date from the side panel, then return to Source Linker and paste it into the birth date field. But it’s my experience that most patrons are not doing this. So more and more profiles in this corner of the Family Tree are being created with only a birth year.
As a result of these profiles with year-only birth dates, and exacerbated by the fact that so many people share the same names, it appears that many more incorrect hints are being accepted since there is less data to help disambiguate between two people (Giuseppe Rossi born 17 Feb 1896 and Giuseppe Rossi born 23 Jul 1896 may look to a patron and to a computer as if it’s a single Giuseppe Rossi born in 1896). So profiles and families end up getting “mangled” together at a seemingly higher rate than before this problem occurred.
What caused this problem and when did it happen? I can’t say for certain, but what I do know is that when these images were first indexed and published several years ago, the full birthdate was showing up just fine in the record. I have some search results exports that I did in October of 2022 and the full birthdate was present. An export that I did in April of 2024 had only the birth year, just like today. So it would appear that the problem image groups got reprocessed somewhere between October 2022 and April 2024 and that there was a bug or a misconfiguration that caused this problem. (There are some image groups that were indexed later and just recently published that are fine so it seems like the problem in the code or the configuration has been corrected. But the records that were already damaged haven’t been fixed).
Is there a workaround for an individual user on an individual birth record? Yes. If you go to the image viewer, and use the side panel to edit the index, you’ll notice that the full date is present (i.e. 11 Oct 1900) but that the date component fields are not all populated. In this example, the year field has 1900 but the Day and Month fields are empty.
If you populate the Day field with 11 and the month field with Ott and save the record, then voilà, the full date appears in all of the places that it was missing!
However, it’s impractical to think that we could educate all of the patrons researching in this area about what to watch for and how to fix it, let alone asking them to fix hundreds of thousands of records that could be fixed by an automated process.
What is the scope of this issue? Possibly 500,000 birth records. I don’t have enough visibility to see exactly. But in doing searches in the province of Udine across a variety of municipalities, including the municipality of Udine, which is by far the largest one, I found only one medium-sized municipality that didn’t have this problem. In all others that I searched, the vast majority of birth records from 1875 to 1900 display only a birth year. That could be upwards of 500,000 birth records. (I also did some random searches in other various cities and provinces across Italy and none that I looked at demonstrated this problem).
Best Answer
-
@ShawnFReid Thank you for your thorough details—they've made my work easier.
Some of my notes:
- Both full dates and single year errors appear randomly on the same DGS.
- The presence of a full date doesn't always depend on whether Month/Day fields are filled in.
- Date parsing is inconsistent, possibly due to date abbreviations in Spanish. Months with identical three-letter beginnings in both languages are more reliably detected and displayed.
I'll share this with engineers to see if the collection can be queued for re-processing.
2
Answers
-
That's awesome, @SerraNola! I really appreciate it. It will be great to see the full intended value restored to this collection!
0










