-
Only registered members can see all the forums - if you've received an invitation to join (it'll be on your My Summary page) please register NOW!
-
It's easier than ever before to check your entries from the 1881 Census - more details here
Geocoding old placenames
- Thread starter Bryman
- Start date
Britjan
LostCousins Star
The practise of limiting availability to academia is the bane of my life. I can understand that it might be "read only" access but there a lot of "savvy" people such as our forum who just want to look at what's available and then ask for specific permission around usage.I note there is a download of historic maps and boundaries at Vision of Britain assuming they are geo-referenced then they could be loaded into the current version of FTAnalyzer. The downside is that they appear to be only available to government or academic institutions
New version would have latest supplied XML file. Ideally the changes are merged into Central XML.
Looking up an old map at Visions of Britain and selecting 19th century it colour codes the counties of England as follows, what is the modern Google equivalents?
Cumberland
Westmoreland
Northumberland
Durham
North Yorkshire
West Yorkshire
East Yorkshire
Lancashire
Cheshire
Derbyshire
Nottinghamshire
Lincolnshire
Staffordshire
Shropshire
Leicestershire
Rutland
Herefordshire
Worcestershire
Warwickshire
Northamptonshire
Huntingdonshire
Cambridgeshire
Norfolk
Suffolk
Essex
Hertfordshire
Bedfordshire
Buckinghamshire
Oxfordshire
Gloucestershire
Berkshire
Wiltshire
Hampshire
Middlesex
Surrey
Kent
Sussex
Somerset
Dorset
Devon
Cornwall
Cumberland
Westmoreland
Northumberland
Durham
North Yorkshire
West Yorkshire
East Yorkshire
Lancashire
Cheshire
Derbyshire
Nottinghamshire
Lincolnshire
Staffordshire
Shropshire
Leicestershire
Rutland
Herefordshire
Worcestershire
Warwickshire
Northamptonshire
Huntingdonshire
Cambridgeshire
Norfolk
Suffolk
Essex
Hertfordshire
Bedfordshire
Buckinghamshire
Oxfordshire
Gloucestershire
Berkshire
Wiltshire
Hampshire
Middlesex
Surrey
Kent
Sussex
Somerset
Dorset
Devon
Cornwall
I'll send you a file.
Actually, I've been thinking some more.
For me, most of the time its not a simple replacement replace A with B type scenario. It's more of the type, if the address is A,B then replace with A,C
e.g.
Speke, Lancashire should be Speke, Merseyside
West Derby, Lancashire should be West Derby, Merseyside
Ulverston, Lancashire should be Ulverston, Cumbria
So, can this be done in the .xml file?
For me, most of the time its not a simple replacement replace A with B type scenario. It's more of the type, if the address is A,B then replace with A,C
e.g.
Speke, Lancashire should be Speke, Merseyside
West Derby, Lancashire should be West Derby, Merseyside
Ulverston, Lancashire should be Ulverston, Cumbria
So, can this be done in the .xml file?
It could but not in its current form
Thanks, sounds promising then.
Beta 4 now has this feature. Still need to see the xml you wrote that didn't work.
Bryman
LostCousins Megastar
More a problem with Google geocoding rather than FTA but can anybody help to explain why it has such difficulty in certain cases. Is there anything that I can do to make the geocoding process more successful?
I have a reference to St Mary's church at Walberton, Sussex, for which the Locations Geocoding Status Report shows . . .
tick? - England, Sussex, Walberton - Partial Match (Google) - Walberton, West Sussex, UK
tick? - England, Sussex, Walberton, St Mary - Partial Match (Google) -St Mary's Binstead, Walberton, West Sussex BN18, UK
! - England, Sussex, Walberton, St Mary the Virgin - Partial Match (Levels) - England, UK
St Mary's at Binstead is more than 1 km due east from St Mary's parish church at the end of Church Lane, Walberton (where I have visited) but seems to be unkown to Google Maps. Giving the name of the church as St Mary the Virgin fools it completely and results in a match for England only, not even Sussex. Fortunately, I know the area in this case and am able to manually reset the location position. However, there are other locations which are not known to me but for which the Google Partial Match is similarly unhelpful.
I am totally confused about some of the various status indicators and what they are supposed to mean. I have looked at the geocoding help on the web but that does not explain. Am I the only person having trouble with this or is anybody else confused too?
I have a reference to St Mary's church at Walberton, Sussex, for which the Locations Geocoding Status Report shows . . .
tick? - England, Sussex, Walberton - Partial Match (Google) - Walberton, West Sussex, UK
tick? - England, Sussex, Walberton, St Mary - Partial Match (Google) -St Mary's Binstead, Walberton, West Sussex BN18, UK
! - England, Sussex, Walberton, St Mary the Virgin - Partial Match (Levels) - England, UK
St Mary's at Binstead is more than 1 km due east from St Mary's parish church at the end of Church Lane, Walberton (where I have visited) but seems to be unkown to Google Maps. Giving the name of the church as St Mary the Virgin fools it completely and results in a match for England only, not even Sussex. Fortunately, I know the area in this case and am able to manually reset the location position. However, there are other locations which are not known to me but for which the Google Partial Match is similarly unhelpful.
I am totally confused about some of the various status indicators and what they are supposed to mean. I have looked at the geocoding help on the web but that does not explain. Am I the only person having trouble with this or is anybody else confused too?
I should write a proper guide. Let's see if I can give you an idea of how the program works and then I can work those ideas into a proper guide.
It takes a location you have entered and looks it up on Google. Google then returns a status code, a list of various possible locations and a string of result types.
The status code is typically OK = means it found a result, ZERO_RESULT = it found no results, or OVER_QUERY_LIMIT meaning that you've had your quota for the day.
When it returns OK it returns a flag to say it was a GOOD or PARTIAL match. A partial match means that it didn't use all the text supplied to find a match, a good match means it was happy that the match used all the text. Note a OK match may not mean it is real match eg: it could have matched against somewhere on the other side of the world.
So what I do is check the result types Google gives back to see if the level of match is similar to what was entered. So if for example you have England, Sussex, Walberton then that has 3 levels so I check if the Google result has at least 3 levels. ie: England, Sussex would be rejected as only 2 levels.
The result is also checked to see if it fits inside the country. eg: if the result is USA and the country checked is England then it fails a bounding box check.
So we have the following status :
Google Matched (small green tick) - Google said it was a good match and the levels were ok and it was in the right country so all seems good.
Partial Match (Google) (tick + ?) - Google said it was a partial match but the levels were ok and it was in the right country so it might be decent
No Match (red x) - Google failed to find any form of match
Outside Country Area (red -) - Match found was outside the boundary of the country
Partial Match (Levels) - Google said it was a match but levels were wrong eg: result type was a county and you asked for a town.
There are also:
Not Searched - location hasn't been geocoded yet
Incorrect (User Marked) - User has marked the location as incorrect
GEDCOM/User Data - User has marked the location as correct (or the lat/long was in the GEDCOM file).
It takes a location you have entered and looks it up on Google. Google then returns a status code, a list of various possible locations and a string of result types.
The status code is typically OK = means it found a result, ZERO_RESULT = it found no results, or OVER_QUERY_LIMIT meaning that you've had your quota for the day.
When it returns OK it returns a flag to say it was a GOOD or PARTIAL match. A partial match means that it didn't use all the text supplied to find a match, a good match means it was happy that the match used all the text. Note a OK match may not mean it is real match eg: it could have matched against somewhere on the other side of the world.
So what I do is check the result types Google gives back to see if the level of match is similar to what was entered. So if for example you have England, Sussex, Walberton then that has 3 levels so I check if the Google result has at least 3 levels. ie: England, Sussex would be rejected as only 2 levels.
The result is also checked to see if it fits inside the country. eg: if the result is USA and the country checked is England then it fails a bounding box check.
So we have the following status :
Google Matched (small green tick) - Google said it was a good match and the levels were ok and it was in the right country so all seems good.
Partial Match (Google) (tick + ?) - Google said it was a partial match but the levels were ok and it was in the right country so it might be decent
No Match (red x) - Google failed to find any form of match
Outside Country Area (red -) - Match found was outside the boundary of the country
Partial Match (Levels) - Google said it was a match but levels were wrong eg: result type was a county and you asked for a town.
There are also:
Not Searched - location hasn't been geocoded yet
Incorrect (User Marked) - User has marked the location as incorrect
GEDCOM/User Data - User has marked the location as correct (or the lat/long was in the GEDCOM file).
Bryman
LostCousins Megastar
Thank you Alexander for going to this much trouble to explain how you perform the geocoding but I think perhaps I am asking too much of you. It may be that Google has just not recorded the lat/long of the insignificant locations that I am interested in.
Walberton Parish Church, built circa 1100/1200, does not seem to be as significant as a small country church nearby at Binstead. That is not your fault.
Recording the map features for geocoding must be a massive task and we should be grateful for what is available.
I also struggled with No Match for the Royal Navy Cemetery at Bermuda but found it on a map without any trouble via a 'normal' Google search!
A little more help/explanation within the Locations Geocoding Status Report would certainly be very helpful, possibly alongside or part of Options and Filters? I think it is the results of Partial Matches which are the most confusing, not understanding the way that the Google processing works.
I have assumed that a status of tick+? probably indicates a better match than anything marked with an ! but that does not always seem to be the case.
There is no simple solution. All results need to be checked as even full matches can be for the wrong location.
Walberton Parish Church, built circa 1100/1200, does not seem to be as significant as a small country church nearby at Binstead. That is not your fault.
Recording the map features for geocoding must be a massive task and we should be grateful for what is available.
I also struggled with No Match for the Royal Navy Cemetery at Bermuda but found it on a map without any trouble via a 'normal' Google search!
A little more help/explanation within the Locations Geocoding Status Report would certainly be very helpful, possibly alongside or part of Options and Filters? I think it is the results of Partial Matches which are the most confusing, not understanding the way that the Google processing works.
I have assumed that a status of tick+? probably indicates a better match than anything marked with an ! but that does not always seem to be the case.
There is no simple solution. All results need to be checked as even full matches can be for the wrong location.
The yellow warning ! is my way of saying the match isn't likely to be that good probably not as good as ones Google said it got right or almost right. It gets a yellow ! when Google gave a result that was "less accurate" than the input data. So if you have a long string eg: England, County, town, street, housename and Google returns a match for town then you get a yellow !. However you'd also get a yellow ! if it returned just a street match as its still less detail than a housename. Now of course not all queries have a housename in the 5th field. Sometimes it could be a less specific thing its text I just can't tell what exactly the form of the address will be. So the yellow ! is my way of saying it could well be right but it might not.
In many cases a partial or level mismatch will be something like England, UK which isn't that specific and is a good candidate for manually editing.
Note that for ALL addresses if you use the Display Geocoded Locations option from the Maps menu on the main form then you can right click on an address and mark it as verified if the Google location looks good enough. This then gives it a bold green tick in a green circle!! Marked as user verified data. In theory you could filter on the things that aren't ticks and work your way through them checking and verifying them. Or even clicking to edit and then moving the marker yourself. This is especially interesting if you use the Bing OS maps or the historic OS maps to find the farm house your ancestor lived in. Note the OS is great for rural locations and not so great in towns/cities.
So you can if you want to spend time on it improve the data Google finds and so improve the various map display functions.
The new Places feature in version 3.3.0.0 due out soon will allow you to browse through a tree of your locations and see them and who lived there plotted on a map. It allows you to discover various branches of your family living close together that aren't related but might provide a clue to where to look for a link.
In many cases a partial or level mismatch will be something like England, UK which isn't that specific and is a good candidate for manually editing.
Note that for ALL addresses if you use the Display Geocoded Locations option from the Maps menu on the main form then you can right click on an address and mark it as verified if the Google location looks good enough. This then gives it a bold green tick in a green circle!! Marked as user verified data. In theory you could filter on the things that aren't ticks and work your way through them checking and verifying them. Or even clicking to edit and then moving the marker yourself. This is especially interesting if you use the Bing OS maps or the historic OS maps to find the farm house your ancestor lived in. Note the OS is great for rural locations and not so great in towns/cities.
So you can if you want to spend time on it improve the data Google finds and so improve the various map display functions.
The new Places feature in version 3.3.0.0 due out soon will allow you to browse through a tree of your locations and see them and who lived there plotted on a map. It allows you to discover various branches of your family living close together that aren't related but might provide a clue to where to look for a link.
Sent.
Bryman
LostCousins Megastar
Is the limit on the number of permitted geocodings each day (mentioned elsewhere in another thread) relative to individual users, or the program (FTA) making the submission?
I just restarted FTA to download the latest version (3.3.2.2) and found that 110 entries in my Geocoding Status Report had been reset to 'not searched'.
I ran the geocoder from the Maps menu and was notified of error 500 (unable to connect with server) after about 30 actions. This happened several times before completion and there was an enforced wait of about 10-15 seconds each time.
When I checked the resulting staus Report I found that one of the locations that I had previously user set had reverted to a more generalised location with ! as status symbol. The location was the parish church of St Lawrence, Hungerford, Berkshire, England which had resolved to just Berkshire, England.
I reset the location pin above the church and saved the result but Google again said that I was over my daily limit of reverse geocodings. I have not used FTA for several days.
I just restarted FTA to download the latest version (3.3.2.2) and found that 110 entries in my Geocoding Status Report had been reset to 'not searched'.
I ran the geocoder from the Maps menu and was notified of error 500 (unable to connect with server) after about 30 actions. This happened several times before completion and there was an enforced wait of about 10-15 seconds each time.
When I checked the resulting staus Report I found that one of the locations that I had previously user set had reverted to a more generalised location with ! as status symbol. The location was the parish church of St Lawrence, Hungerford, Berkshire, England which had resolved to just Berkshire, England.
I reset the location pin above the church and saved the result but Google again said that I was over my daily limit of reverse geocodings. I have not used FTA for several days.