2 replies to this topic

[GJohn]

Hello Renee,

I am interested in the use of the RootsMagic Places fields in the great RM feature "FamilySearch Person Tools." Specifically the interaction of RM & FS so I can make a decision on how I want to utilize the RM place fields.

 

I've been reading forum posts on how to use place names, the dilemmas, many ideas & opinions, good & bad, to the point of confusion. I don't want to rehash this topic. All I want to know is what the algorithms RM & FS software are actually expecting? How strict-forgiving or intelligent is the algorithm?

 

There are four RM place fields.

Which place fields does RM actually send to FSFT when trying to match individuals?
Does any API place conversion or filtering occur before the data is transmitted?

- Place

- Standardized place name (can be auto-populated by RM GeoCode)

- Latitude/Longitude (can be auto-populated by RM GeoCode)

- Abbreviated place name (if any)

 

What type of "ideal" place name does FSFT expect for optimized matching of RM individuals or will make matching worse?

- Modern names (Kansas) vs. Historical name (Kansas Territory)

- The English spelling of names in other countries vs. the native language vs. RM GeoCode mix

     English (North Rhine-Westphalia, Germany)

     German (Nordrhein-Westfalen, Deutschland)

     RM GeoCode (Nordrhein-Westfalen, Germany) mixed?

- Some countries have more than one official language (Switzerland officially has 4 languages)

- What about spellings with special characters like umlauts? German (ü) versus the English substitute (ue)

- What about when the place 4-comma format (city, county, state, country) contain qualifying words like "Smith Township" "Smith Twp." or "Jones County" "Jones Co." in the respective city or county position?

- Are the RM GeoCode database & FSFT equivalent place database the same database?

 

The same place questions if relevant for what is expected by Ancestry.com, MyHeritage, WebHints to optimize their algorithms?

Thank you very much. J

[Renee Zamora]

The FT API will send everything in the Place field exactly as it is. It will not include anything in the Place Details field. It will send a standardized place name to FamilySearch which is a separate level. If it cannot recognize the place name it will standardize it to a level it can recognize. Even if that only means country or state. 

 

Ancestry's API we can send the Place Details as a prefix to the Place name. I do not think they have any place name standard. 

 

On WebHints for MyHeritage and Findmypast there is no transfer of data back and forth. 

[GJohn]

Thank you for the feedback.
Verifying, ONLY the RM Place field is sent to FS, but NOT the RM Standardized Name field, Abbreviated Name field or Lat/Long fields.
Therefore, if I voluntarily choose use a heavily customized Place Name with historical names, other languages or special characters, I may be reducing the effectiveness of the overall RM FamilySearch Person Tool in its interactions with the FS side of the equation.

Does FamilySearch publicly document how the algorithm works on their side, especially what "If it cannot recognize the place name it will standardize it to a level it can recognize"?

FYI: Why I ask? I want to understand how much my conventions for places in RM is interfering with the RM-FamilySearch Person Tools Find Matches. I notice something I know to be the "same place" is often displayed in RED for "poor match". I think its because a FS location might display a modern name for a historic place, while in RM I entered the historic name, sometimes with (dual language) and a [relative timestamp]. After reading many of these forum posts on the "Best RM way" of entering Place names, I realize there is no perfect Place Name convention for many reasons, many beyond RM's control, definitely beyond mine.

An example best demonstrates the Place Names convention I use for an individual who was born & died in the same village, but whose government changed several times over their lifespan.

Birth:   Braunsen, Principality of Waldeck and Pyrmont (Fürstentum Waldeck und Pyrmont), German Confederation (Deutscher Bund) [historical 1815-1867]

Death: Braunsen, Province of Hesse-Nassau (Provinz Hessen-Nassau), Free State of Prussia (Freistaat Preußen), Weimar Republic (Weimarer Republik) [historical 1929-1933]

RM GeoCode Standard Place Name: Braunsen, Hessen, Germany

I often include a separate event with the English spelling of the modern name, contiuing this example, for this Germany village requires 6 fields in the report index (village, municipality, district, admin region, state, country):

Braunsen, Bad Arolsen, Waldeck-Frankenberg, Kassel, Hesse, Germany

I use the Place Details field for name of household, hospital or cemetery and its time relative addresses,

Dashes for same level in report index: "Jon Doe farmstead-RR1" and years later the same place might be "Jon Doe house-123 Country Ln.", for an institution "St. Mary's Hospital-123 Memorial Blvd.",

Comma separated if I want two levels in a report index: "Plot 6-E, Calvary Cemetery-Route 1" or "North Tower-Apt. 1234, Retirementworld-1 Center Dr.",

 

Thanks again. J