[Openstreetmap] maps as graphs

[wabanstar]

If all routes are edges of poygons and all valence 3
nodes and higher are intersections, then which edges
are local, national, etc.? For tyeing datesets into
each other what nodes represent common intersections,
and do edges break polygons when overlayed and why,
what quality issues or data conflicts are represented
and how do you identify common problems and solutions,
and can you automate the solution?
Just a rambling rejoinder.
waban_*
--- Matt Amos <matt at matt-amos.uklinux.net> wrote:
> On Monday 21 March 2005 17:26, Hugh Barnard wrote:
> > Apologies if this is a bit aslant to the current
> topics...Jo (I
> > think, I read the digest) touched
> > on something that will interest me as a 'user':-
> routes (of all
> > kinds) which can be represented
> > as graphs...they tend immediately to be directed
> graphs in cities
> > because of one-way systems etc...also, if one edge
> is (say)
> > pedestrian, then that complete route is
> pedestrian-only etc. etc.
> 
> indeed. graphability gives us a number of very
> useful abilities 
> including shortest path routing with Dijkstra's
> algorithm. 
> 
> > So, a certain amount of the ontology discussion is
> also related to
> > a graphability discussion,
> > don't know how much...
> 
> the graphing stuff is all covered by the
> node-element-area 
> formulation. restricting the intersection of lines
> to occur only at 
> nodes means that all layout, colouring, subdivision,
> etc... problems 
> become graph-theoretic and they've all been solved
> years ago.
> 
> it doesnt solve the ontology issue, which then
> becomes: "what do we 
> call this element? what are its properties?"
> 
> since that is language, country and outlook
> dependent its a Difficult 
> Problem and other people can solve it ;-)
> 
> > Just to declare interest, I've been working on and
> off on a
> > to-be-GPLed rideshare thing which, at present is
> postcode based,
> > but it would produce more matches if it became
> more fine grained in
> > a 'routey' (ha..) sort of way, another reason for
> the post.
> 
> yep. there are several algorithms for counting the
> "common path" of 
> two walks on a graph. this is the same as the
> utility of a ride 
> share, i guess!
> 
> cya,
> 
> matt
> > _______________________________________________
> Openstreetmap mailing list
> Openstreetmap at vr.ucl.ac.uk
>
http://bat.vr.ucl.ac.uk/cgi-bin/mailman/listinfo/openstreetmap
> 


		
__________________________________ 
Do you Yahoo!? 
Yahoo! Small Business - Try our new resources site!
http://smallbusiness.yahoo.com/resources/