Improve FamilySearch Place-Names in Sweden - many places could be automated.
Swedish historical records are for the most part kept by parishes of the Swedish church. Parishes are subdivided into a number of parish regions ("län"). The unstandardized names in FamilySearch's improve place-Names function often follow the pattern of "Parish, Parish regions Län, Sweden", for example
"Örsås,Älvsborgs Län,Sweden" (Örsås being a parish in a parish region named Älvsborg, in Sweden)
This is in fact an old Swedish standard. But in recent years the name standard of the geographical regions of Sweden have changed, so that the standardized name for this would be
"Örsås, Älvsborg, Sweden" (which is indeed a standardized parish name that is suggested by te algorithms). The only difference between the two is a removal of "s Län" (i e "'s region") after "Älvsborg".
In recent years, all of the region names for the parishes of Sweden have been changed from "Älvsborgs län" "Kopparbergs Län" "Stockholms län" to "Älvsborg" "Kopparberg" "Stockholm", which are also the standardized names. Had they been standardized 30 years ago they would have had the "s län" in the names, now they do not.
Hence many older records that need standardized names, already are named by an older standard, and adding standardized names to accompany the ones in the database is a mere formality.
This could easily be automatized so that the database looks for the phrase "s Län" in every entry deemed to be about Sweden, tries to remove it and determines if the result would be a valid parish name for the year in question. If valid, the standardized parish name can be added to the standardized place name field.
Most of these parish names in Sweden are also names of towns or villages, but in effect you always look for the parish name as a Swedish geneologist, since the parish holds all records of births, deaths, marriages etc. The reference is simply one of where to find the record. To further clarify where in a certain parish someone lived according to parish records, you would add that before the standardized parish name, like so:
"Fattighuset, Örsås, Älvsborg, Sweden"
The combination of a parish and a parish region is unique, although the mere parish name without a region is not. But as long as the parish name comes with a specific region and is a valid name for the time period (as is already stated in the FamilySearch standardized parish names list), there is no possibility of misunderstandings or mix ups.
An automation like this, which is very specific, would free volunteer time up to standardizing names that are not as straightforward, such as parish names without regions, which require a little more geographical knowledge and actual fact checking.
Comments
-
Sorry, @twi, but the very mention of "automation" together with "place names" gives me the shivers. You will find MANY threads in the Community about the automated placename algorithm that has created havoc, putting placenames in absolutely the wrong state or country.
No, thanks.
3 -
"Dismal" and "abysmal" are the kinds of adjectives that come to my mind to describe FS's track record with automated processes.
In other words, I agree wholeheartedly with Áine: please, no.
3 -
This is a matter of Sweden changing region's names. All names have been changed from "Älvsborgs län" "Kopparbergs Län" "Stockholms län" to "Älvsborg" "Kopparberg" "Stockholm", which are also the standardized names. Had they been standardized 30 years ago they would have had the "s län" in the names, now they do not. Hence many older records need to be scrubbed from what is a formality using the function of adding standardized place names (which does not remove the written place name), here: https://www.familysearch.org/tree/improve-place-names/
I am not suggesting letting the system do whatever it pleases, I am suggesting machines do a speficic change that humans really do not need to do. I have administered large databases, so I am well familiar with the risks involved in automation.
0 -
An example of the current process: This record is from the US state of Georgia, county of Fulton. Major place - that's where the majority of the city of Atlanta is located. The entire record set is Georgia, since it is the 1933-1998 Death Index for the state. https://www.familysearch.org/ark:/61903/1:1:V4M1-Z1N
The county is indexed twice - once with the name of the state and once without. Trying to attach the record to cousin John's profile a few days ago, I was offered Fulton County, Pennsylvania as a first choice. It's not the oldest in the top 3; it's not alphabetically first; it's not geographically closest.
This is a relatively simple choice - it's a GEORGIA record so the first option should be GEORGIA. It isn't. And if I don't choose, the source linker takes the first option, putting John's death in Pennsylvania.
What you are suggesting is a quantum factor more difficult. Let's get the simple one fixed before we try branching out.
3 -
As another example of the sorts of things that go wrong with anything automated on FS, check out the discussions of source titles (https://community.familysearch.org/en/discussion/comment/538287#Comment_538287).
There are good reasons old-timers like us cringe at the mere thought of "automatic" combined with "FS".
3 -
@twi said
" ... I have administered large databases, so I am well familiar with the risks involved in automation. ... "
Sorry if it sounds like we're insulting your expertise but the risks involved with automating functions in FamilySearch FamilyTree appear to be wholly different from ... well, anything else out there. When I first started here, I took great care to standardise all my placenames for the data that I'd entered. A year or two later - all (it felt like all) those standardisations had gone. That was some sort of bulk process. We never really understood how or why it fouled up. Sorry 😯
4
