* Mapping - existing, Map Facts, and Google Earth

[ColeValleyGirl]

Black Hole of Kolkata or Black Hole of Calcutta? I think the former looks silly, but I might write "Black Hole of Calcutta (in modern day Kolkata)" if necessary

Ah, context. If writing from the English viewpoint, Black Hole of Culcatta. If I was writing from the Indian viewpoint...well, I wouldn't.

Most sources accessible in the UK are probably going to say Black Hole of Calcutta.

[ColinBeaver]

Another option is to use Key Markup Language which is a file format used to display geographic data in an Earth browser such as Google Earth. KML uses a tag-based structure with nested elements and attributes and is based on the XML standard.

It appears this will give the flexible eg you can pick an icon from Google's Earths extensive Library.

I did a post back in 2020

[tparkhill]

I have been through hundreds of Place merges, and I am seeing a pattern that makes me think I should being going a different direction in standardizing. At first, I thought it was just when I Geocode outside FH, but looks like the same problems inside FH.

1. Geocoding does NOT seem to like national abbreviations (e.g. USA, UK, etc.) they seem to not Geocode.
2. Geocoding does NOT seem to like the extra commas in front of the place. (see image below)

As was observed, I was trying to use City, State, County, Country (or equivalents in Europe) and use commas and a space when the parts were not known. e.g.
"Boston, Suffolk, Massachusetts, USA " ; " , Suffolk, Massachusetts, USA" ; etc.

But the Geocoders seem to want . .
Boston, Massachusetts (it prefers to guess the country), or
Suffolk County, Massachusetts
. . and it seems to have it easier if one writes in "County" after the County name.

That is . . . .It feels like the Geocoders were designed to handle what a typical person would do.

Since the Geocoders don't seem to like us putting the same components in the standardized FH "Parts", is there a reason I should not just do what the Geocoders want?

I know in theory I could keep a rigid format for Place, and then in "Standardized" field put in what Geocoders want, but I have many thousands of addresses. As a test, I exported addresses of just births for about 10% of the people I have, and had ~2,500 unique addresses. I don't mean to sound lazy, but the repeated mouse actions in the cleanup activities have started a Carpal Tunnel wrist problem (despite getting an ergonomic mouse, and a variety of heating and cooling things for wrist). Don't think I could standardize 2 sets of all addresses, unless I spread out over a year.

Thanks
Trent

Geocoder and blanks at begining.JPG (27.07 KiB) Viewed 1380 times

[AdrianBruce]

Some comments from me...

... It feels like the Geocoders were designed to handle what a typical person would do ...

Precisely. And therefore it is not impossible that being genealogically clever is counter-productive.

... I thought it was just when I Geocode outside FH, but looks like the same problems inside FH ...

I'm not totally sure what you mean by this. There are 2 ways of Geocoding within FH - the native Geocoder and the PlugIn "Map Life Facts". I know the latter uses a Google Service and the former - doesn't. Generally speaking I prefer to use the PlugIn, especially if it's not practical to audit each result.

... Geocoding does NOT seem to like the extra commas in front of the place ...

Hmm. Both variants of Argyll just geocoded fine for me in the PlugIn. I wouldn't be at all surprised if the native Geocoder did it differently - it does seem variable in its results. All I can say is that - contrary to the preferences of my organising colleagues - I have never used leading commas, not least because I can never get the fingers to remember whether it's space-comma-space, comma-space or space-comma... Nor do I even try to get countries (say) in the same column. Ireland, for instance, can use either a 3 part form (town, county, Ireland) or a 5-part (townland, civil parish, barony, county, Ireland) and I have never found any use for extra commas to pad the 3 out to 5. Sorry - this is one occasion when I stop being Mr Pedantic with a Maths Degree and become Mr Pragmatic trying to fix software at 2 a.m.

... it seems to have it easier if one writes in "County" after the County name ...

Yeah, I would not be surprised - not least because isn't that what real world people do? How else can we expect the geocoding system to understand that "Los Angeles, California, USA" is the county and not the city, other than by telling it? Of course, not everything is as blatant an issue as that but why not get the habit of writing "Co" or "County"? (I have no evidence of whether Co and County perform differently.)

[BillH]

Trent,

I'm not sure what is happening for you, but I have no problems geocoding with country abbreviations and with leading commas. I have 3240 places in my file and never run into any problems with country abbreviations or leading commas.

For example:

Boston, Suffolk County, Massachusetts, USA

geocodes correctly for me.

Another example:

, Berkeley County, West Virginia, USA

geocodes correctly for me.

I have tried both in native FH and in the Map Life Facts plugin.

Bill

[tatewise]

Trent, I am also not sure what is going wrong for you.
Both forms of Argyll, Scotland and , Argyll, Scotland geocode in the FH Maps Window exactly the same.

Please identify the Geocoders you are using and exactly how you use them because there is clearly something amiss.

As you can see, the general response from other FH users is that they don't seem to have as many problems as you and that the Map Life Facts plugin using Google Maps API works extremely well and has no problem with UK or USA.

[AdrianBruce]

... I have 3240 places in my file and never run into any problems with country abbreviations or leading commas. ...
Boston, Suffolk County, Massachusetts, USA ...
, Berkeley County, West Virginia, USA
geocode correctly for me. ...

Interesting Bill - though I have to say that you have the county written the word "County" included. That might help persuade the geocoder where things are.

Certainly I use "USA" and "Co." for all non-British counties, and don't have any current issues using the Map Life Facts plug-in. But I also don't use "UK" - I finish my place-names in the British Isles with England, Scotland, Wales, Ireland etc.... "UK" simply isn't necessary. So I'm not sure how valid my experience is compared to Trent's.

[AdrianBruce]

... Both forms of Argyll, Scotland and , Argyll, Scotland geocode in the FH Maps Window exactly the same. ...

Mind you, didn't we find some odd behaviour the other day with the Native FH Geocoder, where you and I got different geocoding values for apparently the same place-name?

[BillH]

Interesting Bill - though I have to say that you have the county written the word "County" included. That might help persuade the geocoder where things are.

I have quite a few places in these formats:

,, Washington, USA
,,, Country Name or abbreviation
etc.

and I haven't had any problems with them either. I very rarely have a place that won't geocode in the FH geocoder, but those generally geocode just fine in Map Life Facts. I typically have no issues with leading commas or country abbreviations. I spell out County due to the fact that in many states they have townships and counties so it is important to distinguish between them. This is something that Ancestry and FamilySearch don't do and it can be very confusing.

Bill

[AdrianBruce]

... I very rarely have a place that won't geocode in the FH geocoder, but those generally geocode just fine in Map Life Facts. ...

What I tend to find is that when the FH Native Geocoder (i.e. not the Plug-in) does go wrong, then something does encode but it's in a rubbish place - I just checked some American place-names using the Map Life Facts plug-in and discovered that I had a bunch of what I would have thought were perfectly normal American places, which had geocoded in the middle of nowhere. Presumably these were geocoded in the FH Native Geocoder.

I suppose I ought to have been overjoyed to find the rubbish and worked out why they were rubbish but... Alas I was more interested in correcting them. (I do wonder about something that couldn't find "Providence, Providence Co., Rhode Island, USA".... Although there are a number of places called "Providence" in the USA)

[BillH]

As I recall there was a period of time when the "native" geocoder was having problems geocoding just about anything, but that was a few years ago. It has worked pretty well for me for quite awhile now.

Bill

[AdrianBruce]

As I recall there was a period of time when the "native" geocoder was having problems geocoding just about anything, but that was a few years ago. It has worked pretty well for me for quite awhile now.

Bill

That's encouraging, Bill. I suspect that my most idiotic geocodes (Irish place-names on the wrong side of the Atlantic, IIRC) date back a few years ago. They were probably done in the native Geocoder but I can't be totally certain.

I keep trying the native Geocoder - because it's less key-strokes if I just have a single new Place record - but sometimes it's still flakey - one recent example saw a church geocoded against a street of a similar name in a different (English) town. But if other people have different sorts of places, they may be perfectly satisfied with the native Geocoder.

[Little.auk]

In the case of Sudeley, Gloucester, England, I found that Google maps went there and Wikipedia has co-ordinates for Sudeley. The Maps Life Facts plug-in gets there but the native FH Geocoder doesn't - at first glance it's in the middle of the Gloucester suburbs (possibly the result of the middle name???) but when you zoom right in, it's on Sudeley Way in the suburbs of Gloucester. So it's not totally incorrect. Just useless!

Useless perhaps, but I think it is the most accurate match for the Place Name given! Because FH and Streetmap have found a Sudeley reference in Gloucester (the CITY).

If you correct the Place Name to use the correct County name i.e. Sudeley, Gloucestershire, England FH returns a point close to Sudeley Castle, near Winchcombe.

So - a perfect example of the pitfalls of automatic geocoding, which is why I keep it firmly switched OFF!

Sudeley as a place (village town or city) does not exist, it is only a Parish in Gloucestershire --- the Sudeley Parish Church of St. Mary's is in the grounds of Sudeley Castle. Also within the parish is Sudeley Hill.

Why this extra information? - Just to demonstrate the value of geocoding for yourself and what you can learn about a place while doing it, in my case courtesy of Google Maps.

While on the subject of Geocoding -- For places in England, Scotland and Wales can I recommend the Gazetteer of British Place Names at https://gazetteer.org.uk/ not only is it a useful reference, but it also has a downloadable Gazetteer in csv format, which I have found invaluable.

[Little.auk]

All this talk of merging place names makes me feel very nervous!

Quoting from the FH Knowledge Base - "Genealogical best practice recommends that you document the (place) name as it was recorded in the source that refers to it, i.e. an historical name" .

Over the generations my maternal family place of residence has changed county three times, so, following best practice, I have records showing each of these as perfectly valid place names for the records concerned.

Merging these to the current place name would result in two sets of valid place names being changed, to show a "county" that didn't exist when the events occurred.

The recommended way of dealing with these is to use the 'Standardised Name'

[Gowermick]

All this talk of merging place names makes me feel very nervous!

Quoting from the FH Knowledge Base - "Genealogical best practice recommends that you document the (place) name as it was recorded in the source that refers to it, i.e. an historical name" .

Peter,
Best Practice doesn't mean obligatory.

My stumbling block is Warrington, which moves back and forth between Lancashire and Chesire
I settled on Lancashire. As long as I can place a mark on the map, in the correct place I'm happy.

Make FH work for you, not the other way round. I do a lot of things that some may shudder at, but it works for me, and others can follow what I've done, that's the important bit.
So do what you think is right for you!
BTW There is nothing stopping me having both Warringtons in FH

[Little.auk]

But I tried the variants I have for Sudeley:
Sudeley, Gloucester, England
Sudeley, Gloucestershire, England
Sudley, Gloucester, England
. . . and a few others that had Sudeley Manor in Place 1.
But Google did not seem to find it.

I refer you to my earlier post - You will not find a place (village, town or city) called Sudeley - If it ever existed, it doesn't now, except as a parish, East of Cheltenham and South of Winchcombe.

[Little.auk]

Best Practice doesn't mean obligatory.

Quite true, but I did say 'recommends'.

I prefer to observe it, so I have a number of places with two or three entries, including a few where a place is known by two "different" names, e.g Sturton Le Steeple, Nottinghamshire (aka Sturton cum Fenton), and Mallerstang, Cumbria (aka Moraston) - neither "aka" names appear on maps.

[Gowermick]

I refer you to my earlier post - You will not find a place (village, town or city) called Sudeley - If it ever existed, it doesn't now, except as a parish, East of Cheltenham and South of Winchcombe.

Well, it is shown on my FH map as a village (openstreetmaps)! Type Sudeley in location search bar, and it goes straight there

[ColeValleyGirl]

But is that the viiage of charlton abbotts which i believe is the biggest village in sudely civil parish (which has only existed since 1935).sudely cp was formed from (among others) sudely manor parish, which formed the northern part of the new civil parish and contained sudely castle and sudely manor. No village.

https://visionofbritain.org.uk/unit/10392196#tab02

[Little.auk]

1. Geocoding does NOT seem to like national abbreviations (e.g. USA, UK, etc.) they seem to not Geocode.

There is an International Standard - ISO 3166 - that defines 2 and 3 character codes for Countries. For the United States these are US and USA. I have just one American place in my family tree, Tampa, Florida, USA - which geocoded without any problem.

The abbreviation UK is a different matter, the United Kingdom code is not UK, but GB and GBR - which the standard assigns to the UK's full name of the "United Kingdom of Great Britain and Northern Ireland". I have never tried using GB or GBR abbreviations

[Little.auk]

I refer you to my earlier post - You will not find a place (village, town or city) called Sudeley - If it ever existed, it doesn't now, except as a parish, East of Cheltenham and South of Winchcombe.

Well, it is shown on my FH map as a village (openstreetmaps)! Type Sudeley in location search bar, and it goes straight there

Read what I wrote in the post I referred to, or check an O.S. map!

Openstreetmaps is taking you to the centre of the Sudeley Parish (as does Google Maps) this is quite near the very tiny village of Charlton Abbots. Sudeley Parish Exists, Sudeley Castle Exist, Sudeley village does not.

[AdrianBruce]

... There is an International Standard - ISO 3166 - that defines 2 and 3 character codes for Countries. ... I have never tried using GB or GBR abbreviations

I suspect I could spend ages trying to break geocoding in all sorts of ways in a bid to reverse engineer its recognition algorithms. I can say that (a) CA is the two-letter code for Canada, (b) that in the native FH Geocoder, "London, CA", which clearly means "London, Ontario, Canada", gets geocoded to London, California - which is in Tulare County and had a population of 1,869 in 2010. As distinct from the 422k of the Canadian version

So while that geocoding facility might recognise the ISO codes, ISO 3166 seems down the priority list...

[AdrianBruce]

... Over the generations my maternal family place of residence has changed county three times, ...

For a long time, where counties changed over time, I always entered the place under its pre-1974 details. Thus, "Manchester, Lancashire, England" even in the 21st century. Various things, however, kept nudging me to question this practice and when it came to entering the places where my parents had died, the emotional impact was such that I simply couldn't ignore the data on their DCs. Their deaths were after Cheshire had been split into Cheshire West and Cheshire East (administratively) - if I'd carried on with my old practice, I'd have entered the places as "Cheshire" - but it seemed wrong to enter something that wasn't on their source, and wasn't really "right", so Cheshire East and Cheshire West it was. (There doesn't seem any need to enter Standard Place Names for the earlier "Cheshire" place-names - perhaps with the word Cheshire still being on the map, the old "Cheshire" places still appear to geocode fine today.)

As an aside, there are places where I enter place-names that aren't on the source. The classic case is events taking place in the railway town of Crewe. BMD certificates usually record the address of the event as Monks Coppenhall, since that's what officialdom regarded the area as. I prefer not to confuse people and I enter the place for the event as Crewe. (Basically, so far as I can see, Crewe was the name used by the world at large, Crewe Municipal Borough was the jurisdiction, but the area covered by Crewe MB was defined by the old Township, later Civil Parish(?), of Monks Coppenhall. Complicated.)

[Little.auk]

! can say that (a) CA is the two-letter code for Canada, (b) that in the native FH Geocoder, "London, CA", which clearly means "London, Ontario, Canada", gets geocoded to London, California - which is in Tulare County and had a population of 1,869 in 2010. As distinct from the 422k of the Canadian version

So while that geocoding facility might recognise the ISO codes, ISO 3166 seems down the priority list...

Sorry Adrian, to the geocoder that isn't "clear" at all!

It highlights the problem with using abbreviations instead of Names - Codes are only unique in the class they belong to!

In your example, although CA is the Country Code for Canada, it is also the State Code for California! The geocoder has no way of knowing which one you mean.

Which one it returns depends on the algorithms of the geocoder, in this instance, "London, CA" looks most like a US State Address, as this format is commonly used in the USA to differentiate places having the same name.

[AdrianBruce]

... "London, CA", which clearly means "London, Ontario, Canada", gets geocoded to London, California ...

Sorry Adrian, to the geocoder that isn't "clear" at all! ...

Urrgh - sorry Peter. I had thought that splattering the "wink" icon was sufficient to indicate that I was being sarcastic about "London, CA" clearly meaning "London, Ontario, Canada". Apologies. Your point about the double meaning of "CA" is exactly what I failed to successfully get across.

I do try to encode my placenames as clearly as possible - which is why (being serious) I dislike intensely the use of, say, "Germany" before 1871, as it is unclear what is meant by it. Is Austria in or out, for instance? (Yes, the usual answer varies by date, but the ambiguity persists)