Making FamilySearch Source Titles Easier to Understand with the AI Research Assistant

ChrisPetersen

The following list of sources was copied by clicking on the Banker box (“Sources used by AI Research Assistant during chat”) and then using the copy function. It includes the default source titles and links to the records.
The titles that appear in bold below were created using ChatGPT based on the information indexed by FamilySearch. These improved titles are clearer and easier to understand. They also help prevent incorrect record linking and make FamilySearch more accurate.

Currently, the titles that FamilySearch generates include only the name of the person the record is attached to and the title of the collection. As a result, many different records end up with identical titles, such as:
“Wendelini Rothenberger, ‘Deutschland, Baden, Erzbistum Freiburg, katholische Kirchenbücher, 1463–1931’.”

Additionally, changing a title must be done separately for each individual to whom the record is attached, which is time-consuming.

With a small improvement, the AI Research Assistant could easily generate clearer, more descriptive source titles automatically—producing better outcomes and reducing errors across the FamilySearch platform.

FamilySearch Sources (Already Attached to Wendelin Rothenberger MQBB-RB2)

Wendelinus Rothenberger, „Deutschland, Baden, Erzbistum Freiburg, katholische Kirchenbücher, 1463-1931“ | https://familysearch.org/ark:/61903/1:1:QP3C-HP7F or Germany, Baden, Catholic Church Records — Marriage of Wendelinus Rothenberger and Anna Maria Treber, 16 Feb 1779, Oberhausen; bride’s father Nicolai Treber

Joannis Rothenberger, „Deutschland, Baden, Erzbistum Freiburg, katholische Kirchenbücher, 1463-1931“ | https://familysearch.org/ark:/61903/1:1:QP3C-YL1Por Germany, Baden, Catholic Church Records — Baptism of Anna Maria Rothenberger, 6 Apr 1760, Oberhausen; daughter of Joannis Rothenberger & Maria Barbara

Wendelini Rotenberger, „Deutschland, Baden, Erzbistum Freiburg, katholische Kirchenbücher, 1463-1931“ | https://familysearch.org/ark:/61903/1:1:QP3Z-3TBR or Germany, Baden, Catholic Church Records — Marriage of Joannes Adamius Rotenberger and Maria Catharina Cindemann, 8 Nov 1784, Oberhausen; groom’s father Wendelini Rotenberger, bride’s father Cindemann

Wendelini Ratenberger, „Deutschland, Baden, Erzbistum Freiburg, katholische Kirchenbücher, 1463-1931“ | https://familysearch.org/ark:/61903/1:1:WXR6-57ZM or Germany, Baden, Catholic Church Records — Burial of Anna Catharina Ratenberger, 9 Apr 1800, Oberhausen; daughter of Wendelini Ratenberger & Anna Maria Treber

Wendelinus Rothenberger, „Deutschland, Baden, Erzbistum Freiburg, katholische Kirchenbücher, 1463-1931“ | https://familysearch.org/ark:/61903/1:1:WXRF-31PZor Germany, Baden, Catholic Church Records — Burial of Wendelinus Rothenberger, 25 Nov 1807, Oberhausen; age 86 (born about 1721)

Joannis Wendelini Rothenberger, „Deutschland, Baden, Erzbistum Freiburg, katholische Kirchenbücher, 1463-1931“ | https://familysearch.org/ark:/61903/1:1:WXRJ-3SPZor Germany, Baden, Catholic Church Records — Marriage of Georgum Jacobum Mayer and Eva Catharina Rothenberger, 5 Aug 1788, Oberhausen; groom’s parents Nicolai Mayer & Margarethae Treber; bride’s parents Joannis Wendelini Rothenberger & Mariae Barbarae Breith

Wendelini Rothenberger, „Deutschland, Baden, Erzbistum Freiburg, katholische Kirchenbücher, 1463-1931“ |https://familysearch.org/ark:/61903/1:1:WXRN-MVZM or Germany, Baden, Catholic Church Records — Marriage of Martinum Weser and Catharina Barbara Rothenberger, 17 Aug 1790, Oberhausen; groom’s parents Joannis Weser & Maria Ursula Alt; bride’s parents Wendelini Rothenberger & Maria Barbara Breidhaupt

Adrian Bruce1

@ChrisPetersen said:

Currently, the titles that FamilySearch generates include only the name of the person the record is attached to and the title of the collection. As a result, many different records end up with identical titles …

Indeed. If I recall correctly, the source record titles used to be more informative but are now very terse and ambiguous. I'd love to know why the change took place (or even be contradicted if necessary).

I definitely agree improvement is needed. However, I have two concerns about this precise suggestion…

Firstly if a (pattern recognition) AI solution is to be used (rather than a straight algorithmic approach) - what is the certainty that the AI will give reproducible results (i.e. the same each time) without hallucinations?

Secondly, given that index source records refer to the persona within the real world source (i.e. to the bride, the groom, the bride's father, the bride's mother, etc, etc) the titles of the index source records need to include the persona, so we need a robust method of ensuring the correct persona within the source record is being described in the title. (This process had issues even with manual indexing because if a persona wasn't present in the original - e.g. a mother missing on a baptism - there was a possibility that a dummy persona was created named "/", instead of omitting the persona completely).

Again, let me emphasise that I wholly agree with what is suggested - it's just that I have some concerns about the how.

ChrisPetersen

Eva Catharina Rothenberger: 27 Sources with Identical Titles

Eva Catharina Rothenberger (26 September 1765 – 23 September 1816, ID: GWFY-KDQ)

Eva Catharina Rothenberger currently has 34 sources attached to her profile in FamilySearch. These sources come from three different collections.

• 27 of the sources are from the same collection (Deutschland, Baden, Erzbistum Freiburg, katholische Kirchenbücher, 1463–1931).
• Each of these 27 records carries the same collection title, but the name in front of the title varies (e.g., Rothenberger, Rotenberger, Rothenberg, Rother, etc.).
• Unfortunately, the records do not indicate the type of record (such as baptism, marriage, or burial), which makes it difficult to determine their exact context.
• The repeated identical titles obscure the true value of the sources.

I am sharing this here to highlight the issue and to see if others in the community have encountered similar situations. Has anyone found a way to better identify these record?

📂 Grouped by Collection

1. Germany, Births and Baptisms, 1558–1898

• Eva Katharina Rothenberger in entry for Franz Martin Mayer

2. Deutschland, Baden, Erzbistum Freiburg, katholische Kirchenbücher, 1463–1931

• Eva Catharina Rothenberger
• Eva Catharina Rotenberger
• Eva Catharina Rothen
• Eva Catharina Rotenberger
• Eva Catharina Rothenberger
• Eva Catharina Rothenberger
• Eva Catharina Rothenberger
• Eva Catharina Rothenberger
• Eva Catharina Rothenberger
• Eva Catharina Rothenberger
• Eva Catharina Rotenberger
• Eva Catharina Rothenberg
• Eva Katharina Rothenberger
• Eva Catharina Rotenberger
• Eva Cathar
• Eva Catha Pater
• Eva Catha Rothenberger
• Eva Catha Rotenberger
• Eva Catha Rotenberger
• Eva Catharina Rothenberger
• Evae Catharinae Ratherberger
• Eva Katharina Rother
• Elisabetha Bothenberger
• Eva Catharina Rotsebermer
• Eva Catharina Rothenberger
• Eva Katharina Rothenberger
• Evae Catharinae Rotenberger

(Total: 27 entries)

3. Deutschland, Baden, Erzbistum Freiburg, katholische Kirchenbücher, 1678–1930

• Eintrag für Valentin Maier
• Eintrag für Eva Katharina Maier
• Eintrag für Georgum Jacobum Mayer
• Eintrag für Andreas Maier

(Total: 4 entries)

MandyShaw1

I don't know if it will help, but I have done some analysis on all GWFY-KDQ's profile and relationship sources (too big to paste here, see Source analysis for GWFY-KDQ.xlsx).

There is less duplication in the images than I expected, with the citations varying considerably (and showing a wide range of record dates); the source titles basically just match the not very specific persona titles, which is of course what the Source Linker will put in place.

ChrisPetersen

Thank you for sharing your source analysis file—very interesting work. It provides a helpful overview of the sources attached to GWFY-KDQ and highlights how much variation there is in the citations and record dates.

I’m curious about your process. Could you explain how you created the analysis and how you use it in your research?

Also, I would appreciate clarification on a couple of the column headings in your spreadsheet:

• Collection
• Collection Type

What do these represent, and how do they relate to the sources in FamilySearch?

Thank you!

MandyShaw1

https://community.familysearch.org/en/discussion/comment/612975#Comment_612975

Thanks Chris.

I have an analysis database which I populate with the data of FT profiles of interest using the FS read-only APIs and a bit of RootsMagic. The aim mostly is to try to cast a different light on data-related problems that may arise here on Community.

The Record Collections referred to are FS' way of grouping historical records, and thus the personas these records contain, and can for example be filtered on when searching Records. Each Collection has a unique id, and a type indicating (in theory) the type of records it contains.