GeoNames Forum

Hi Kappuz

Thanks, for letting us know.
I looked into it and found an inconsitancy for administrative division codes that confused the geocoding algorithm.
I have rerun the geocoder with certainly better results.

Cheers

Marc

Hi David

I have checked all US places against the timezone polyons used for the timezone web service and updated all erroneous timezones.

Marc

Hi

Thanks, I have inserted them in the DB, without lat/lng at the moment. I will geocode them later.
I have also added Correo Argentino as one of the data sources :
http://www.correoargentino.com.ar/

Cheers

Marc

Hallo Casper

Eine fertige Lösung haben wir noch nicht. Eine Idee ist die administrative divisions zu verwenden, dort wo eine touristische Region gewisse admin regionen umfasst. Diskutiert haben wir das zum Beispiel in Spanien (Inseln) oder auch Indonesien. In beiden Fällen könnte man die touristischen Regionen als Teilmenge von admin Regionen darstellen.
Touristische Regionen sind ein Beispiel, wo die normale administrative Division nicht die einzige Regionenaufteilung sind und wir parallele Hierarchien und Beziehungen modellieren und abbilden sollten.

Gruss

Marc

Hi Sami

Some of them are certainly populated places. The data is from the US geological survey who are using the type 'populated place' for them.

There was a similiar discussion some time ago, where we changed everything the USGS has not designated as PPL as road :
http://forum.geonames.org/gforum/posts/list/143.page

Even among those that don't have population number we find some real cities : http://www.geonames.org/4754808/cross-junction.html

I don't know of an easy way to filter them out. We would have to cross check against population, wikipedia, zip codes and TIGER.

Marc

Hi

Great resource. The license of the dataset is not very clear, so I hope we may use it.

Marc

Sami

I have increased your userlevel and Christope has fixed the PPLX for Paris 04.

Marc

date attributes are missing so far in the GeoNames data model. It is not yet possible to model changes of features over time.
I cannot say whether it is better to keep them or not, normally I would say they should be deleted. Unless they continue to have some kind of landmark function.

Marc

Hi David

There is something wrong with the timezone stored in the database. I will fix it with the values the timezone web service returns. It should be fixed this week.

Cheers

Marc

Hi Emj

Interesting question. I think what you refer to as sociological divisions are often called neighbourhood. Zillow, the US real estate service, have released neihbourhood polyongs for the US :
http://geonames.wordpress.com/2008/01/24/neighbourhood-reverse-geocoding/

For other counties I don't know of any similar data collections.

Marc

Hi Julian

Maybe it would be better to not have a admincode2 at all for London the capital as it is spanning all London Boroughs.

Marc

Hallo Ralf

Der ws server ist ein anderer Server, bei den modifikationen über das Webinterface werden die modifikationen mittels Message Service übermittelt. Diesen Update habe ich aber direkt auf der DB ohne den Message Service gemacht und musste deshalb die neuen Daten kopieren. Was jetzt auch gemacht ist.

Gruss

Marc

Danke Ralf, ich habe die DB mit deinen Daten aktualisiert.

Marc

Hi Fred

I have sent you the link to the documentation via email. I don't want to officially release the insert/update/move API right now. I prefer to run it in some beta testing for a while to figure out what further consitency checks we need. I think the risk of errors and duplicates over a web service batch API is bigger than for manual updates.

Cheers

Marc

Hi Julian

The PPLA2 has only been added as a tag, it is not yet an official feature code. I don't know whether it makes sense to add this as a new feature code. There is a general problem with the PPLA feature code since PPLC often also serve as PPLA and we cannot model this information, with PPLA2 it will even get worse. With the tagging system we can add as many tags to a toponym as we like. This might be a more flexible solution for properties like PPLA2.

I agree the situation with London is not yet satisfactory, at least it has not gotten worse with the introduction of the countries as ADM1.

We have been discussing about adding GOVs as ADM2 in this thread :
http://groups.google.com/group/geonames/browse_thread/thread/ecc2154d8d0ba8f1

There the list of GOVs in the UK (Scottland, Wales and Norther Ireland don't have GOVs and count as one GOV) :
East Midlands
East of England
London
North East
North West
South East
South West
West Midlands
Yorkshire & Humberside

And here the list of London Boroughs :
City of London
Barking and Dagenham
Barnet
Bexley
Brent
Bromley
Camden
Croydon
Ealing
Enfield
Greenwich
Hackney
Hammersmith and Fulham
Haringey
Harrow
Havering
Hillingdon
Hounslow
Islington
Kensington and Chelsea
Kingston upon Thames
Lambeth
Lewisham
Merton
Newham
Redbridge
Richmond upon Thames
Southwark
Sutton
Tower Hamlets
Waltham Forest
Wandsworth
Westminster

The capital of London we have in the GeoNames database is a feature class 'P' wheras the administrative divisions are a feature class 'A'. In most countries we have inserted two records to distinguish between 'A' and 'P' functions of a geographical entity. This is another disadvantage of the strict feature type classification system.

We didn't decided not to model the GOVs at all. They are (except for London) not very well known and might confuse most users of the GeoNames database.
In the case of London I have the feeling that the Boroughs are also (besides their administrative function) something like a PPLX for the PPL London. What do you think?

Marc

Hi Justin

Sure, I can help you with the data and the code for address reverse geocoding in the form of a consulting service. My email is marc@geonames.org
Postgres has much more sophisticated spatial support then MySQL, though it is possible to run it with MySQL.

Cheers

Marc

Hi Styler

What do you consider a 'real' city? The data for the US is from the US Geological Survey.

Cheers

Marc

Hallo Sebo

Grundsätzlich stimmt deine Hierarchie.

Hamburg ist als kreisfreie Stadt verwaltungstechnisch ein Speziallfall und als ADM3 eingetragen : http://ws.geonames.org/search?q=Hamburg&country=DE&isNameRequired=true&style=Full&featureClass=A&featureCode=ADM3

Ich weiss nicht, was für Hamburg mehr Sinn macht. ADM3 oder ADM4 oder vielleicht sogar beides um den Doppelfunktionen gerecht zu werden.

Gruss

Marc

Hi Justin

The data for the address reverse geocoding is not included in the dump. It would bee too large (several GB) and addresses are not the main topic of this project. What is included in the dump are the editable places.

The data for the address reverse geocoding is from the US Census and it needs a spatially enabled database to achieve acceptable performance. We are using postgres/postgis.

Cheers

Marc

Hi Fred

I have fixed the problems with Utrecht, Lisse, Dokkum and Krimpen aan den IJssel. Thanks a lot for finding them and pointing them out.

There is a web service API to update records, it is 'beta' and nowhere linked to in the documentation. It does not have a command line interface, but it would be possible to build one on top of it.

Cheers

Marc