GeoNames Forum

I just spent a day or so researching this issue -- then came here and found this discussion. One of these days I'll get things in the right order.

By my calculations there are 8328 German records for which the admin level codes are incorrect. I also believe that the records to the ADM1 level are correct. At first I thought the problem was that "Landkreis" records are at the ADM3 level when they should be at ADM2 and the ADM4 under the Landkreis records should be at ADM3. But then I find that some German states use a "Regierungsbezirk" layer which really is the ADM2 that is above the "Landkreis" level. Note that not all states use this structure so for those states there is no ADM2 level. All in all the data appear to be correct. However this begs the question: Is the data modelled correctly?

Perhaps we should start a discussion about how to best model the data bases on usage. For the most part I think that, in most cases admin levels are used to start/restrict searches. So, for example, one might want to find all the lakes in Weld County, Colorado, USA and so start with that restriction. In a sense the admin records are used to traverse up and down a hierarchy.

Perhaps a structure that uses up and down pointers (a b-tree approach for the technically minded) might be in order here.

The information I received is:

Please be kind to find the list of second administrative divisions
(by the first administrative divisions) on the following link:
http://hr.wikipedia.org/wiki/Dodatak:Popis_gradova_i_op%C4%87ina_u_RH_po_%C5%BEupanijama

as well as from Croatia\'s official gazette:
http://narodne-novine.nn.hr/clanci/sluzbeni/127788.html

In addition you may want to use the EUDict to translate some of the text.
http://www.eudict.com/

I was going to check the list against the list of administrative divisions. In the first document you are looking for entries in the "Općine:" sections.

I have a user that has pointed out that several of the 2nd level administrative divisions in Croatia are missing. He has provided me with a website that lists all the 2nd level divisions but I can't find a way to insert them myself, so I assume this is a bootstrap process. If so, how should the data be submitted for inclusion and what format is recommended/requested?

After reviewing the administrative level 1 codes, I assume that all countries should have a 00 code that represents objects not found within the other administrative divisions. While I disagree with this strategy (IMHO a "null" should be used), I want to make sure that it is correct.

If so I will provide a list of countries that have 00 encoded admin1 attributes but no ADM1 record with a 00 code.

One more question. Are each of the administrative codes supposed to build upon the previous code. Fore example a record might have value2 for country_code, admin1, admin2 and admin3, but you should not find a record that has values for country_code,admin1 and admin3 without admin2 being specified. Is that correct?

as a follow on, each adminX field specified should have a corresponding feature_code='ADMX' record, with the same adminX value constraints specified above. Is that correct?

If I understand this correctly then, an entry for a city might have an admin1 code if there are admin1 codes in the system for the country in which the city resides. To find the admin1 code name I take the country,admin1 combo from the city and look up the record that has that country and admin1 code as well as feature_class='A' and feature_code='ADM1'.

If that is right then so far so good and I can ask my question, which is: Should there be any records where the feature_class='A' and the feature_code='ADM1' and the admin1 fields is empty?

I was doing a referential integrity check on a complete set of data from Dec 8, 2009 and found 79 records ADM1 records that have no admin1 value.

Brazil has an invalid language code listed as pr_BRR. note the extra "R". I can not find anyplace to change this value. Does anyone else know where the country language codes are stored?

The time zone data has Asia/Kathmandu
The Goedata lists Asia/Kathmandu 77 times
The Geodata lists Asia/Katmandu 1158 times.

I think that it should be changed in the Geodata to match the timezone data. How does one go about contributing this?