GeoNames Forum

Please see my original report in this thread, where I explained what's not attached and what seems like it ought to be.

I clearly point out the parent-child relationships of two GeoTree nodes that, although I am showing you this in GeoTree form, is evident in the Web services queries for hierarchy and children.

I'm not sure what else I could tell you to help you understand. It seems painfully clear to me that I have illustrated the nature of the problem, in the most plain terms with examples, and yet on the other side of this equation you are making very cryptic or blunt statements expecting me to have intricate, arcane knowledge of how the repair ought to be performed.

Whether something has changed or not (according to my records it has), I have plainly explained what's missing by the two examples.

I'm sorry if it's not enough. It's the best I could do.

I honestly don't understand.

We pay a yearly fee to access the Web service.

The cities that previously were correctly attached are now no longer attached.

I don't know where I'm supposed to post an inquiry into this.

I don't have any idea what it means to say "you can build it differently using the web service."

I think you are assuming far too much familiarity on my own part, with your own internal process of doing updates.

If this is a self-service operation, at least point me to the documents where I can help get this fixed?

> 1. I don't see any adm2 relations on this wikipedia page

I didn't offer that as a complete solution; if I had a complete solution I wouldn't post here in the first place...

> 2. I am interested in real data not html wikipedia pages. For instance boundary polygons for adm2 divisions.

I am not a geographic data expert, I am a paying customer for your licensed product. I shouldn't have to delve into this level of detail to report a problem and invite a solution. I can't exactly demand you provide a solution (I realize that's not in the terms of your service), but on the other hand should I as a customer expect a demand that I be the resource who provides technically complete data?

> 3. feel free to update the adm2 yourself: http://www.geonames.org/manual.html - Who, if not you? When, if not now?

Who if not me? How about someone who really understands the technicalities of geonames, its data structure, mission, and so on. Maybe you? Maybe a happy volunteer who does have these answers (which I don't)?

When if not now? You tell me...

I understand this is a proto-volunteer project in some respects. Perhaps I have misread your various comments but in the past it seemed as though you cared about the quality of data and paying customers. Maybe I've misunderstood the nature of your organization.

m.

Well, I'm a little puzzled here.

If you're looking for a source for cities, perhaps (though who knows) this may be of help:

http://en.wikipedia.org/wiki/List_of_cities_in_Australia#New_South_Wales

But again, the Geonames database itself contains many many cities previously under New South Wales, which are now orphaned for no apparent reason. Why not simply reconnect those, if they have been improperly severed?

Checking the GeoTree view of New South Wales, Australia (2155400), there are just nine (9) children.

However, previously, there were many cities underneath 2155400.

In fact even now, if you search for Broulee (2173824), you will find a city that still refers to New South Wales as its parent.

Compare the following GeoTrees:

1. http://geotree.geonames.org/2155400/ (shows 9 cities beneath NSW)
2. http://geotree.geonames.org/2173824/ (shows many cities beneath NSW)

Also the "children" service for 2155400 returns just 9 items.

Why have so many towns and cities been removed from GeoTree #1, and why aren't they returned in the children service for 2155400 -- ?

Thanks!

The island of "Île Bizard" (GeoNameID: 5980072) does not appear as a direct child of Quebec, nor do any of the other of hundreds of islands that refer to Quebec as their parent (in their own local Geoname tree information).

Is this a mistake? Why don't the islands appear as children, even though they refer to Quebec as their parent?

marc wrote:

The difference is that VI is modeled as ADMD (administrative division) whereas VG is a PCLD (dependent political entity).

Maybe it would make sense to model the US dependencies as PCLD too. What do you think?

Best

Marc

Just for the record, I have changed the U.S. Virgin Islands to PCLD.

I haven't made a broader effort to change U.S. dependencies in general to that classification, but I think (based on my understanding that the U.S. Virgin Islands carries its own flag and has its own ISO-issued country code, and so on) that the PCLD designation makes more intuitive sense to me than ADMD in this case.

marc wrote:

"Bay of Islands" is not a top level region, it is a child of Northland. When information is missing as in this case then the hierarchy service returns the next known ancestor.

Admin divisions need an admin code for the assignment to work. The admin code for a admin division cannot be modified with the interface. If no code is present the assignment will add the code for 'undefined' to a code that was already 'undefined' and thus the error message.
I have added the FIPS code 'F6' to Northland. Tasman does not have a FIPS code and I have added the geonameid as it's code.

Marc

Okay, but every change you've made here seems to be making things much, much worse. For example now, there is only one child under Northland (previously although Northland was missing under New Zealand, it contained many toponyms).

I don't really understand how to fix this as New Zealand is really getting into very bad shape now...

I have pointed out an error and problem in trying to update data. I am quite willing to assist with updating data, but if the interface won't save or produces errors, I am at a loss as to what to do.

I have added further information to the "New Zealand" issue under the Toponyms discussion, but have not received a reply in several days.

Any chance of a reply?

Another clear example where I cannot determine how to fix the data by editing directly...

1. Go to "Bay of Islands" (county)
2. Edit the data
3. Attempt to change the 1st administrative region (below "New Zealand") from unassigned, to the correct region ("Northland")
4. Click Save

This fails with the error "error while saving: no changes"

Tried in Safari and Firefox, same results in both cases.

I was able to make some corrections in individual records, but attempts to change parentage failed, producing the following error(s):

"error while saving: no changes"

(example: after successfully changing the name "Wanganui" to "Manawatu-Wanganui," attempting to change the 1st order administrative region in the crumb trail for this record from "Wanganui" to "Manawatu-Wanganui" failed with the above error.

I have found what appears, unless I am confused about something here, to be a very serious split between two sets of data. I have taken a screen shot (attached) showing the two views of New Zealand (both of which are substantially wrong, although for different reasons).

Here is the top-level region/subdivision list that the Statistics bureau of New Zealand uses:

Auckland Region
Bay of Plenty Region
Canterbury Region
Chatham Islands District
Gisborne Region
Hawke's Bay Region (misspelled as Hawkes Bay under NZ: 2186224)
Manawatu-Wanganui Region (misspelled as Wanganui under NZ: 2186224)
Marlborough Region
Nelson Region
Northland Region (missing under NZ: 2186224, present under 218597

Otago Region
Southland Region
Taranaki Region
Tasman Region (missing under NZ: 2186224)
Waikato Region
Wellington Region
West Coast Region
?? Whangarei (improperly placed as a region under NZ 2186224, but it is actually a District under the non-present "Northland" region)
?? Te Anau Downs (incorrectly attached directly under 2185978, but it is actually within the Southland region of New Zealand)

I would enter these if I knew how, but right now it looks like the data is already a bit mixed up in Geonames, and until I become more familiar with editing and re-parenting data, I don't want to mess up something if I'm not really familiar with it.

I will be glad to help though, if you could just direct me how to deal with this situation.

marc wrote:

I don't think 'Bay of Islands' is a top level region of New Zealand, and therefore it should not be a direct child of New Zealand.

you can help adding missing information yourself:
http://www.geonames.org/manual.html

Although it is not a top level region, it is given by the New Zealand census bureau as a district that is listed parallel to the top level regions. Perhaps, as it associates a group of islands, it is not geographically large enough to constitute a political region but, as a geographic association, it is not a subset of a particular region and therefore should be a direct child of New Zealand.

This would tend to follow your precedent of grouping by geographic block rather than cultural or political association (as in the case of Turkey); it is also tacitly endorsed by the New Zealand official bureau of Statistics. See:

http://www.stats.govt.nz/census/2001-census-data/territorial-authority-leaflets-by-region.htm

I will see what I can do to correct this issue within the Geonames data, but this is my first attempt at making corrections so I will let you know what I do, in this thread, after I complete the changes.

Thanks!

I will be glad to add some information, if I can.

However, I request clarification on this point:

If "Bay of Islands" is not a direct child of New Zealand, what is it a child of? Is the aim for geographic subdivision (as with placing Turkey under Europe) or is it for political subdivision?

I don't see how to navigate to "Bay of Islands" if its parent is listed as New Zealand, but somehow it is not a child of New Zealand. This makes no sense to me.

I am happy to do some updates for Geonames but I don't want to proceed in a state of confusion about this point.

A further problem:

Chatham Islands itself now no longer appears to show any children at all (where once it had many, including Bay of Islands):

http://ws.geonames.org/children?geonameId=4033013

Thank you! That indeed solves one part of the problem.

However, "Bay of Islands" still cannot be found directly under (as a child of) "New Zealand," even though now the parent issue is fixed for Bay of Islands directly.

Shouldn't it be added as a child of New Zealand too?

According to my retrieval of the hierarchical data for GeonameID: 2189493, Chatham Islands is the direct parent of Bay of Islands.

Furthermore, Bay of Islands is not a direct child of New Zealand.

See:

http://ws.geonames.org/hierarchy?geonameId=2189493

marc wrote:

It is a geographical decision, not a political one. The geographical border between Europe and Asia is the Bosporus. According to this 97% of Turkey is in Asia and only 3% in Europe.

GeoNames lists two populated places with the name 'Dalyan', one in Çanakkale Procince and the other in Muğla:
http://www.geonames.org/317678/dalyan.html
http://www.geonames.org/317681/dalyan.html

Best

Marc

Thank you. A perfectly reasonable answer!

I have no personal knowledge of the Island regions of New Zealand, but the following was brought to my attention:

In Geonames, under "Oceania > New Zealand", one child is listed as "Chatham Islands." Still further, underneath that, is "Bay of Islands."

I don't know if there is a political reason, but there seems to be no geographic basis for this decision. A quick query of Google Maps shows that Chatham Islands are on the northeast coast of New Zealand, while Bay of Islands are on the other side of New Zealand proper, far off the eastern coast.

Some people living in Bay of Islands claim that this is totally wrong, and should be corrected. Perhaps they need to be listed side-by-side instead of one as a child of the other -- I don't know, and could not find any Wikipedia entry or other information to verify. I only found one case where "Bay of Islands" is below "Chatham Islands" but I think they are using your data.

Can you confirm?

The person making this argument to me also provided the following corrections for several locations in Turkey:

--

OK! USE GOOGLE MAPS !!
BUT READ THE RESPECTIVE LISTING'S GEOGRAPHICAL INFORMATION FIRST
TO AVOID GRAVE ERRORS!

FOR EXAMPLE :
DALYAN IS NOT IN ÇANAKKALE PROVINCE(AEGEAN) BUT
IS IN MUGLA PROVINCE(SW MEDITERRANEAN)!!!!!!

KALKAN İS NOT IN KUTAHYA P$ROVINCE(CENTRAL ANATOLIA)BUT
IN ANTALYA PROVINCE 5SW MEDITERRANEAN)!!!!

I HOPE THIS INFORMATION HELPS YOU!
SO PLEASE MAKE THE NECESSARY CORRECTIONS!

AND FOR GOD'S SAKE

TURKEY IS NOT IN ASIA BUT
IS IN E U R O P E !!!!