Did you see the automatic suggestions for place names?

Question

Hi WikiTreers,

We just implemented a long-awaited feature: automatic suggestions for place names.

This is currently only available when editing profiles. When it seems to be working pretty well there we'll add it elsewhere.

This utilizes FamilySearch Place Names database (thank you, FamilySearch!) and includes a cool research feature. You will see a little pin icon next to suggestions. If you click the pin you'll go to FamilySearch place names research tool, e.g.
https://familysearch.org/research/places/place/10564035

For advanced members, I want to talk about how this relates to some pretty esoteric but important style and policy issues.

The perfect place name may not be among the suggestions. It's important to note that you can enter names as you have in the past.

FamilySearch's database is not the exact, perfect, ideal database that we would have designed and built for WikiTree if we had unlimited resources to design and build such things.

Our style rule is to enter the location name that the person themselves would have used at the time of an event. This is an almost-impossible ideal for a variety of reasons, but our rule has a good rationale that we don't need to go into here.

FamilySearch's database is almost perfect for us because it has historical place names and they're available in a variety of languages. But FamilySearch intends the language setting for the user. Users can access place names in their own language. There is no consistent connection between place names and the predominant local language (let alone the predominant local language at specific times in history).

We decided to use FamilySearch's language setting to improve the suggestions for our purposes as much as possible.

We did this by adding a Language field to profiles. You can now select from a couple dozen modern languages to indicate the language the person spoke. If you don't do this, the place name suggestions will be in English.

I've been worried that the place name suggestions and the Language selection would cause a lot of confusion.

Our style rule on place names is confusing enough, when you get into the weeds of it. Now we've added a Language field that is supposed to be ... what? The language they spoke at birth or death? What they spoke at home, or what local officials wrote? It doesn't actually matter. Currently this language field is just used for the place name suggestions. Hopefully it won't cause too much controversy and we can develop a style rule for using the language field if and when we need it.

You'll notice a big yellow alert that comes up along with suggestions. I'm not sure if this will stay permanently. I don't like the feel of it. For now it gives us the opportunity to explain some of this.

In the new year we can tweak this based on how it's working out.

Merry Christmas!

Chris

Comments

I was in the middle of updating my Stanley profiles when I noticed the change - I saw that the birth and death dates now appear before the birth and death locations. I was surprised, but didn't miss a beat!

I also saw that the message box that says I don't have to use the location autocomplete menu now appears at the upper left, and is not in the way (Firefox browser) - another useful improvement.

Well done, Chris & the Cracker-Jaxx (my new name for our Awesome Tech Team)!!

 

---

I noticed the new message box while editing a profile. I liked the fact that it was out of the way. However, I expect that there will be new concerns from people who say that they can't see white text on a "yellow" background.

---

The flouting massage box is a huge improvement.

Many Thanks

---

The floating box is a huge improvement.  Thank you so much Chris!

---

Hi Chris

Please help 

On this page https://www.wikitree.com/wiki/Language_Selection

If you use the language setting, the place name suggestions will appear in that language. If you don't select a language, the suggestions will be in English.

The language button on South African profiles (at least the ones I was working on)  does not have Afrikaans as an option. If i go with the "other" option everything is still in English. 

I hope you can :fix" this 

It would make the entering of place names so much more fun and a lot less work, an typos far less.

If Not Thanks anyway for trying  

 

---

Sorry, Ronel. This is FamilySearch's database we're using. If they add an Afrikaans option we can add it.

---

Hello Chris,

  there is Czech language in https://familysearch.org/research/places/ but not in wikitree Language list. Can You pleas add it?

---

Will do. Thanks, Stanislav!

---

I would love to see this feature added to new profiles and the form for adding/editing marriages.  According to my error report, I can't seem to spell Connecticut correctly :)

---

Stanislav, Czech is now an option. I'm sorry for the delay.

Answer 1

Thanks, Chris. I have begun going through our Italian ancestors adding the Language on already finished profiles. When I find one with the name entered in English, I attempt to change it. What I found unusual was the name of the churches in parenthesis in the middle of some suggestions for the same town. Is this common in the FamilySearch database or perhaps something unique in Alia, Sicilia? The church name would indicate the place of baptism.

Is there some way that the language spoken by the person profiled can show up in the main profile page? I would like to add Italian to all my Italian profiles, but is it worth it to go back and do that if it does not show on the profile page?

Thanks again!

Sharon Centanne.

Comments

I have edited a couple dozen profiles to reflect the language as Italian. I then save the profile.  When I go back, the selection is gone. Is this only a temporary selection? I would love it to be permanent.

Also, can you set up a bot that wold automatically put the language of each person profiled based on their place of birth, maybe using categories? Any rare instances of people speaking a non-native language would be easier to change by hand than every profile in town.

 

Thanks again!

Sharon Centanne

Answer 2

Working great!

Answer 3

Chris, I'm of two minds on this one. It does save typing, but there are other ways to do that (I use copy/paste a LOT, plus Firefox's autofill). My biggest objection, however, is that FamilySearch's database is too inaccurate!

Example: Most of the entries for West Virginia (USA) counties are wrong. Their database keeps suggesting that a particular county is actually a town or district within another county! Unless you know the geography, you won't know that it's wrong. Generally, the correct suggestion is in the list, but it rarely is the first suggestion, and sometimes not even the third or fourth. This is a place where use of the term "County" in the place name helps greatly, and their database doesn't include the term at all.

I see in the other responses that my problem with county names is not unique, as Eva Ekeblad noted -- I think her example is in Sweden(?). From her response, I gather that inclusion of the term "Parish" might be really helpful, as well.

Also, changing boundaries is another problem. If you aren't careful about the history of a place, it's easy to use the wrong suggestion from their database. I'm most familiar with that problem in the USA, but I know it also occurs in Europe, & probably just about everywhere. For example, use of "Germany" or "Austria" when "Holy Roman Empire" would be the accurate term.

Another problem is the use of "British Colonial America" which WikiTree has decided to deprecate. It's standard for FamilySearch. (That's where I found it originally, & accidentally started the discussion that came to the conclusion not to use it on WikiTree.)

It needs to be very plain that these are suggestions only, & not the final word about what is a valid entry. This is especially important for the casual WikiTree'er or the newbie, who won't ever see this discussion.

And if this list overrides the browser autofill (I haven't used it enough yet to tell), then I (very regretfully) vote to discontinue this new feature. It's the proverbial straw that broke the camel's back...

Yes, it might prevent typos, but I think the drawbacks outweigh the advantages in this case.

Comments

Yes, Kitty, my examples are from Sweden. It's not exactly that i want he WORD parish included, rather that I find it very unhelpful when the parish name is omitted, because the parish is the basic unit for finding sources in Sweden.

I would be very interested in hearing where else this type of omission of the crucial part of the location name might occur.

Sweden cannot be the only country where there are 1) villages/farms with the same name in many different places, or 2) villages with the same name as a parish elsewhere. I guess there may be other types of confusing similarities in other countries - I'm just trying to think of this in a more abstract fashion. I'm not addressing the problems of historical changes, I know too little of that topic.

Now, I can understand if somebody just knows that their ancestor came from Sverige, or just from Småland, Sverige. So these very general locations have to be acceptable to the system, and also exist as suggestions. I can also understand situations where it has been passed down that an ancestor came from Karstorp (a farm/village name) and nothing more. This is a difficulty, of course - since there are Karstorps in many different places. The system would have to accept Karstorp, Sweden as an entry. I'm not so sure locations of this type should come up as suggestions, though. Or, well, OK, if it is OK that nobody is ever going to be able to follow the paper trail further back, then OK.

What I can most assuredly NOT understand is when somebody has scoured the ministerial books of a cluster of parishes for every ancestor, sibling and cousin, way back long before 1700, and transcribed them using only the village name and the province, omitting the parish name, although they must have been well aware that the parish is the source of sources. The FamilySearch location database contains a lot of this type of material for Sweden, I suppose for other countries as well. It is, after all, a repository of volunteer work, isn't it? But it needs a thorough cleanup to be useful.

---

I agree. The FamilySearch database is not adequate for my entries in Banat either.

---

Eva, if Sweden is anything like the USA, using the term "Parish" could be very helpful to distinguish what portion of the location is the parish name, as opposed to the village or province. Example: if someone wrote "Mason, West Virginia, USA", did they mean Mason the town, or Mason the county? On FamilySearch, there is no way to tell because the term "County" is not used in their database. There is a Mason town, which is in Mason County. Mason County is over 1100 sq km, most of which is completely rural farmland. As of the 2010 census, Mason the town had less than 1000 inhabitants. Mason the county had over 27,000 inhabitants, & about 18,000 did not live within the boundaries of any town or village. West Virginia does not have formal parishes or townships, so it is very common for the most specific location possible to be just the county.

As for ignoring the parish name when transcribing village names & provinces, I will give someone the benefit of the doubt & assume they had no idea that the parish name would be important to future researchers. This sounds like an American bias at work -- aren't parishes (as political entities) generally a result of a state religion? While there are parishes in the USA, they are strictly religious, and while they may keep/have kept their own records of birth, baptism, marriage & death, those records are generally much harder to find & search than those kept by political divisions in this country. We have the added challenge of first finding out which religion to search for a record...

---

That's true, the strong linkage of the parish to population records is an effect of the state church. It may well be an American bias to overlook the importance of parish names for finding sources.

However, what gets to me, when I struggle with sourcing some old imports to WikiTree, is how the initial transcribers of the records have delved into some really obscure old books, which must have been clearly organized parish by parish on the films - and then omitting the parish name when entering the location data, in spite of having having it right in front of them.

I guess this would be like writing "Mason, West Virginia, USA" if Mason the town was in a different county from Mason the County. I guss those situations must exist in the US as well.

---

Yes, there are situations in the United States where a single name can refer to both a town and a county, and they are in completely different places. My favorite example of this (because it's one I can remember; there are plenty of other examples) is "Decatur" in the U.S. state of Tennessee. The town of Decatur, Tennessee, is more than 200 miles from Decatur County, Tennessee.

Answer 4

After reading through answers and comments to this and related posts I get a sense of support and acceptance, with some minor issues, from members who work mostly with US and UK profiles but hardly anybody working with non-English speaking areas seems to be happy.

• Language: The selection is very limited, Czech for instance which could be useful for the Czech Roots Project is not available and all pre-independence locations default to "Bohemia, Austria".
• Historic locations are extremely limited: I have yet to find anything in Germany with the correct attribution prior to 1860 which puts the place of death for Charlemagne into Prussia; Italy appears to have only modern regions, Bologna exists only in the Emilia-Romagna, Italia, and Rome only in Lazio, Italia; the same appears to be true for Spain.
• US centric listings in the form town, county, state, country, neglecting Swedish parishes which appear to be important there, or other regional customs.

On balance the supporting up-votes for approving answers outweigh those for critical comments probably reflecting the make-up of WikiTree users. For me, however, this seems to be a step away from becoming truly international and instead an embrace of US customs.

Comments

To put this a little more structurally, as opposed to my carping about the Swedish parishes: are there other countries where FamilySearch tends to omit the unit that is most important as a "source of sources"?

---

In my area of research (Bohemia) it is not so much that sources are obscured as we have regional archives containing all extant church books with good indexes allowing easy identification of the parishes where records were kept. It is the historic inaccuracy that makes the database useless. To say that everything before independence was in Bohemia, Austria, implying that Bohemia was a part of Austria is akin to saying that Sweden was a part of Denmark during the Kalmar Union, or the United Kingdom a part of Hannover. For Bohemia it is only true for the time after 1806, before that (since 1526) the Archduke of Austria was also King of Bohemia in personal union.

---

You mean that all individual persons are indexed?
Or do you mean that all villages are indexed as to parish?

It is not difficult to find out which parishes in Sweden are in which county or province, and not difficult to find out which archive the preserved records are in either. But indexes for persons in SVAR start basically in 1880.

This is not a problem for me in my own research - it is a problem only when I try to source profiles from old GEDCOMs :-)

---

Person indexes vary widely from parish to parish but can be as early as the beginning of the 17th century, marriage indexes unfortunately often contain only the groom's name. To their great credit the staff of the regional archive for South Bohemia is gradually putting together digital indexes that connect directly to the image of the church book entry, and for both sexes for marriages.

---

That is great, the digitizing of indexes, I mean.

The situation with marriage records only giving the name of the groom I recognize all too well. If the bad luck continues she will only be mentioned as "wife of" in her death record.

Mothers are also seldom mentioned in the older birth records. Indeed, it is very common that children out of wedlock also are given as child of their father, with no name for the mother. One might have to search for the mother in the court records (which I have no skill with).

---

BTW, "person indexes" in olden times might be what I call household records? There are parishes when they go back pretty far. And other parishes where they start VERY late. There are also taxation records - mostly listing only the heads of household.

---

In Bohemia church records were the official state records so there is no other way to find anything about the women as tax and land records often only mention the male. Out of wedlock children, however, give us the opposite problem as they are typically only listed with the mother unless they got legitimated by subsequent marriage.

As to "person indexes": some parish priests apparently took it upon themselves to keep indexes early on way before they were mandated. Those indexes exist sometimes as separate books, sometimes as appendices at the end of a book of records. They typically list the name, volume and page number, later also the year and sometimes even the full date of a birth, marriage, or death.

---

Later on out of wedlock children are typically listed only with the mother, so the first time I encountered children listed with only the father I was surprised. Even later there was the possibility for the mother to give birth anonymously - their name went unrecorded; they obviously had to go away from home to have the child. I have never had occasion to research a case like that.

Bless the parish priests who made indexes without being told to! In the towns of Sweden there will sometimes be that kind of indexes - without them it can be very frustrating to research townspeople.

Answer 5

I gave every answer an upvote on this. Great steps being made. However, I would also like to see the fields added to so that "their time and place" is properly adressed.

As an example - when I tried to add Swellendam, Grootvadersbos, de Caep de Goede Hoop, Dutch Cape Colony [South Africa] to this profile, it automatically brought up this The Grootvadersbosch Estate, Swellendam, Cape Province, South Africa which is incorrect for that time and place.

a) Grootvadersbosch was not an estate then but a farm

b) It was in the district of Swellendam that was in

c) The Dutch Cape Colony with the name of Caep de Goede Hoop which was

d) not the province of the Cape [Cape Province - only came into being in the 20th century] as did

e) the political and national entity that became known as the current South Africa [only after 1902].

As far as place of death goes:

He died in Schoemansdal, Zuid-Afrikaansche Rebubliek [South Africa].

The Transvaal did not exist then as such, though it was also known as Transvaal - what existed during the time of his death was the political and national entity of the Zuid-Afrikaansche Rebubliek - there have been numerous G2G discussions on this and I agree with Paul [Bech] and Michael [Gabbard on this (see this thread).

Comments

 I have several births in Mafeteng, and Tsikoane in  Basutoland  which was part of the Cape Colony for a short period in the late 19th C and then became a Crown Colony. Basutoland doesn't exist in the data base as far I can see . I just  get Lesotho  which it didn't become until independence in 1966

Answer 6

Good Dayy,  WikiTree Members

I am finding many good reasons for this, "meant to helpful tool."

I like the objectives for the use of this tool.

However, I like others have discovered that the suggestive list does not work accurately for many regions, historical time periods and locations that have already accurate and established descriptions on WikiTree and not to found on the imported list.( Or other descriptions continuing to be added to WikiTree.)

For me this is OK.  It is an "option." I am working "around" and "with" the list and making certain I am adding correct descriptions of the period and locale and that they are sourced and include a historical source reference to the description if it is necessary.

There are helpful opportunities made possible from using the "suggesting" list.

My experience with the location identifiers from Familysearch are that they are sometimes poorly researched and are not as accurate and of the same quality of many of those already to found on WikiTree. (My opinion and based from my specific research and subject matter I work with.)  

Ideas for improvement.

1.  Add and enhance the existing script message to boldly and expressively state that it is a list of suggestions intended to be helpful and not intended to replace historically established and accurate circa descriptions. IT is an option.

2. Add and combine all our WikiTree established descriptions to the suggestion list as options to select from.

I am currently helping someone with a recent significant sized gedcom import to WikiTree with locations that appear to have been automatically chosen from the " option list" that are not accurate and the accurate descriptions are already established and available on WikiTree. I am not certain how to address that gedom import process other than I am helping to correct and edit the locations to each profile one at a time.

C'est Bon

Answer 7

If we are a wiki with aspirations of being a global family tree, why are we glomming off another U.S.-centric website's poorly maintained locations  database with no transparent process for making corrections or improvements?

Answer 8

Thank you for more improvements, Chris!  I'm sure it will be an improvement for many new genealogists.  My huge, huge, pet peeve is that County or Co. is not included for US place names.  I hate that on Family Search trees and dislike it intensely on imports from Ancestry.com and other sites. I think that it is ambiguous: i.e., Midland, Michigan.  Where is that?  Is it the city or the county?  Or Hamilton, Ohio?  Is that the city in Butler Co. or is it Hamilton Co., Ohio?  I'm sure many other wonderful WikiTreers could add place names that are confusing.  I see that it is a suggestion to use and really like that some of the place names have dates.  Hopefully no one will be annoyed by my use of County or Co.

Comments

I fully agree regarding the need to include "County" in the names of U.S. counties, Kathy! Please don't refrain from pestering people about this.  :-)

Answer 9

I'm very glad to see this giant step towards location normalization! And FamilySearch's database is a very reasonable choice.

Are the database place IDs being stored when a suggestion is chosen?

(It would be nice to see a future version filter or at least order the suggestions using relevant event dates.)

Comments

Hans, the 2 last ancestors managed by you in Önsbach have the following time and location:

26 Apr 1809 in Önsbach Achern, Ortenaukreis, Baden-Württemberg, Germany
7 Feb 1842 in Önsbach Achern, Ortenaukreis, Baden-Württemberg, Germany

This would be a reasonable choice if we would have a policy to use the current location and English instead of "using place names in native languages and using the names that people at the time used, even if they now no longer exist." Önsbach became a district of Achern in the 1970's, before that it was an independent village; the Ortenaukreis only exists since 1973; and Baden-Württemberg since 1952. Whether anything like Germany beyond a rough regional attribution existed in 1809 or 1842 is highly debatable.

Given the lengthy debates in G2G about correct locations in time it is absolutely astonishing how quickly we are ready to throw all that over board and call it "normalization"!

---

Well, this is a splendid case study regarding my need for tools. ;-)

(My own laziness about researching the history of place names notwithstanding.)

I hope you don't think "normalization" for me means aligning names at the expense of date context. Quite the opposite: I dream of an ideal with a hierarchical, temporal places gazetteer that would have guided me to the correct names on these records while also keeping them "linkable" for analysis and mapping.

---

I'm working almost completely outside the US with my genealogy and, unfortunately, getting more and more frustrated with the very database we have chosen to provide us with place names. In principle I'm all for such a tool but I do think at present the one we are using is just creating more problems than it solves. Here and there are suggestions popping up to create our own database which of course would be a huge undertaking but may be necessary.

---

I more-or-less proposed this here a while back, but I think Chris is wise to bootstrap with an external source first. Unfortunately, I cannot find a means to propose updates to FamilySearch. :-(

---

Btw, you may be stuck with your line in Önsbach: Prior to 1792 it belonged to the parish of Fautenbach and their records start in 1724, older records were destroyed in a fire in 1734.

Answer 10

Chris,

I really like the recent change to move the instructions to a pop-up box at the upper right, rather than in-line.

As a further enhancement, would it be possible to remove duplicated entries from the list of values for the placenames? Many seem to be repeated several times.

Comments

The duplications are in the FamilySearch database of place names.

---

Yes, I appreciate that, but that doesn't mean that they can't be filtered before being displayed.

Answer 11

Must say I detest the thing, keeps popping up in my face when I'm doing profile improvements, and never has the correct place name in it.  Also the pop-up box showing on the right is a pain.  

Please!  This is NOT facebook or other social media, let's keep it that way!  This is a collaborative work in progress.

Answer 12

First, thank you for putting this in place, even though it's far from perfect.

I mostly work Belgian and French profiles, and here's what I have to say:

- The good point, as far as France is concerned, is that it should limit the biggest location errors. No more "Nevers, Puy-de-Dôme, Clermont-Ferrand" or "La Rochelle, Isère, Rhône-Alpes". The suggestions, as far as I can see, actually place locations in what is (or was at some point) the correct departement and region.

Unfortunately, it won't help with anachronic place names such as people born in 1200 in Yvelines, or Kings sacred in "Reims, Marne, Champagne-Ardennes"! Ancien Régime (before 1790) provinces are not covered at all (except Paris). Also, departements that changed names between 1790 and now don't seem to be covered either. There is one pretty cool feature though, it can manage the change when Seine-et-Oise and Seine were replaced by different departements, Paris, Hauts de Seine, Val d'Oise etc (1968) so you can see directly which to use if you're recording an event before or after 1968.

Also, I avoid entering regions when I record French locations, because they're no use (departement names are unique in France, and therefore town+departement+France is sufficient to identify a place). Unfortunately, the tool systematically suggests them. That's likely to create more errors, since French regions changed less than 2 years ago: one more reason not to use them.

- Belgium: the tool had no there are no historic locations - nothing to address place names before 1830, before Belgium was Belgium. Post 1830, it is not language consistent: it mixes up location names in French and Dutch: for instance "Mechelen, Mechelen, province d'Anvers, Belgique" which is not good; it is either "Malines, Anvers, Belgique" or "Mechelen, Antwerpen, België" not a mixture of both.

Bottom line: when entering profiles, I still have to Wikipedia each place name to find its historic province. Never mind, I actually enjoy doing this!

Answer 13

has it already been mentioned that the automatic place names don't work for Marriage locations? (I didn't read the whole thread)

Comments

Hi Dennis. We were waiting until everything was fully settled before applying it elsewhere. It's probably good to go so it should be added there soon.

Answer 14

It pains me to enter a Czech Republic place name when there was no such entity in the 1850s Austrian Empire. BUT I want my Family database to group all the people who lived in a geographic point descibed by latitude and longitude, no matter what century. This doesn't seem so tough from a programming standpoint.I value the Wikitree policy of describing place AT THE TIME in the current language. I will update my files.