Re: Correct hierarchy for Melilla, Spain
Marc Wick
Ceuta and Melilla are very special cases. They are autonomous Spanish
cities located in Africa. In a way they are like autonomous regions
(adm1) in other ways they are more like municipios (adm3). For the other
Spanish ADM1 we use the fips code, but fips did not assign a code to
Ceuta or Melilla. They have a ISO code which would put them into the
category of ADM2.
Whatever we do somebody will complain. The question is how should it be
modeled?
The easiest would be to have it 4 times (adm1,adm2,adm3 and as city).
Everything else will mean working with holes and missing entities
between father and child, but if we add it four times you can be sure
that people will ask why the same city exists four times in the database.
Best
Marc
Martin Sarsale wrote:
> Dear Javier - Spain Geonames Ambassador:
>
> I've found there're two 'Melillas' in spain which are non related:
>
> http://www.geonames.org/6362988/melilla.html first-order administrative division
> http://www.geonames.org/2513947/melilla.html seat of first-order
> administrative division
>
> I think the 2nd one (2513947) should be child of the first one
> (6362988). Do you agree? Do you think you could update it?
>
> Thanks!
>
Marc Wick
cannot easily be modelled by a general model. It would be nice to have a
simple global model that covers all special cases that can happen in all
countries. I fear this will be impossible, because it also depends on
the angle you are looking at it. For these special cases there simply is
no 'correct' solution that satisifies all wishes.
Marc
Martin Sarsale wrote:
>> Hi Martin
>>
>> Ceuta and Melilla are very special cases. They are autonomous Spanish
>> cities located in Africa. In a way they are like autonomous regions
>> (adm1) in other ways they are more like municipios (adm3). For the other
>> Spanish ADM1 we use the fips code, but fips did not assign a code to
>> Ceuta or Melilla. They have a ISO code which would put them into the
>> category of ADM2.
>>
>> Whatever we do somebody will complain. The question is how should it be
>> modeled?
>>
>> The easiest would be to have it 4 times (adm1,adm2,adm3 and as city).
>> Everything else will mean working with holes and missing entities
>> between father and child, but if we add it four times you can be sure
>> that people will ask why the same city exists four times in the database.
>
>
> I'm not sure what the correct solution would be, but it really breaks
> my preprocessing code to construct the hierarchy: I end up with two
> places named 'melilla' and IMHO it's obvious that one should be
> 'nested' inside the other.
>
> But maybe this is just because I misunderstood the algorithm to
> recreate the hierarchy. Some documentation about it would be
> ---really--- appreciated; I cannot find the Melilla city in geotree so
> I'm puzzled!