Brad Neuhauser <brad.neuhau...@gmail.com> writes: > On Wed, Oct 20, 2010 at 11:06 AM, M∡rtin Koppenhoefer <dieterdre...@gmail.com> > wrote: > > 2010/10/20 Peter Budny <pet...@gatech.edu>: > > 2. Defining how "important" a city is (and thus, how big its label on > > the map should be) is a tricky thing to do. Population is certainly a > > large factor, but how do you define this? The City of Atlanta is the > > #33 most populous city in the US, with 540,000 people, but the Atlanta > > metropolitan area is #9 with 5,475,000 people and is the largest metro > > area in 800 miles. > > good you mention this, there are similar cases in Europe. E.g. > Stuttgart counts 601.646 inhabitants, but the metropolitan area has > 5.3 Million ranking 12th in Europe before Munich (5.2 Million > met.area, 1.33 million inhabitants city) > http://de.wikipedia.org/wiki/Metropolregion# > Die_gr.C3.B6.C3.9Ften_Metropolregionen_Europas > > We already have had similar discussions on the German list, where the > result was to add as much detail as you can to help the rendering > application choose the one they are interested in. > > The resulting matrix is here (in German, sorry): > http://wiki.openstreetmap.org/wiki/DE:Anzeige_von_St%C3%A4dten#Deutschland > > The legend: > > HS = Hauptstadt --- capital > MR = Metropolregion --- metropolitan region > F = Flughafen --- airport > H = Überseehafen --- harbour with overseas traffic > B = Bahnknotenpunkt --- important railway intersection > OZ = Oberzentrum --- main regional centre (?) > MZ = Mittelzentrum --- medium regional centre (?) > Uni = University, one x per 10 000 students > > another approach from the same page is titled "dominance" where > dominance expresses the distance to the next "higher" (in terms of > importance / population) place. The higher (in terms of distance) the > more dominant. > > This serves to determine which names to show and which to omit. (in > scarse areas you would want to see also smaller places, but in > concentrated areas you will have to omit also big cities in favour of > even bigger (or more important according to a scheme like the above > described one) ones.
That dominance method seems pretty cool, and similar to what I was thinking (although probably a little better thought-out). Compute a "score" that represents how important a place is, based on a combination of lots of factors, and start labeling from the top down as space permits. > Very interesting, Martin, thanks for sharing. Two things I'm unclear > on is > 1) whether this matrix influences what "place" level the city > gets (or maybe these are all large enough cities, so this point > doesn't matter) I think that if this dominance scheme (or something like it) were used, place= would become irrelevant except to mark the actual legal status of something besides what admin_level= tells us. That is, in the US where cities, towns, and townships are all admin_level=8, place= could tell us which is which, but otherwise would have no bearing on the labeling. > and 2) what is the mechanism to get this info to the > label renderer, because I don't see anything on the "place" node. I suppose we could just tag all place markers with a city_dominance_score= tag... or we could just add a step in the rendering pipeline to calculate it automatically from various datasets and some rules. That would make it much easier to generate varying maps based on what the user considers "important", e.g. driving directions vs sightseeing vs geo-political maps. With a single static tag, there isn't as much flexibility. However, I'm not sure whether it's worth making the rendering process more complicated just to add that. -- Peter Budny \ Georgia Tech \ CS PhD student \ _______________________________________________ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
