- From: Jos De_Roo <jos.deroo@agfa.com>
- Date: Tue, 20 Jan 2004 22:50:27 +0100
- To: "pat hayes <phayes" <phayes@ihmc.us>
- Cc: Austin Tate <a.tate@ed.ac.uk>, public-sws-ig@w3.org, public-sws-ig-request@w3.org
PatH: >>PatH: >>[...] >>> BTW, the same abstraction works very well for space. Traditional >>> spatial models start with points (where you click the digitizing pen >>> on the map) then go to (oriented) line-segments (pair of points) then >>> to end-linked sequences (paths) then to closed paths (end=beginning >>> and no crossings) which define (oriented) 2-d regions. These all >>> build up very nicely as finite structures on points. You can even go >>> to 3-d and higher, using ideas from homology theory, by stitching >>> together 2-d cells into 2-d surfaces and defining closure by >>> cancellation of oriented edges. All of which suggests that this is >>> indeed a very robust (and certainly very simple) framework. >> >>That sounds great Pat - can you recommend any URI pointers ?? >>Thanks in advance :) > > Still no URI, but I found the relevant part of my paper copy of the > 1997 USIGS Data Model. In sum, its basically this: > > Locations are described by Geometric-Spatial-Elements > G-S-Es are Surfaces or Points or Lines or Volumes > A Point in a Surface is a Surface-Point... > <various subclassifications of surface types, eg fan-surface, etc.> > > Point is the locus of a Node > Line is the locus of an Edge > Surface may be covered by Faces > .... <etc. relating actual topography items to topological idealizations> > > Now: Node, Edge, Ring, Face and Shell are all Topological-Elements, > and they are related as follows: > Node is identified by a Point (with coordinates) > Edge has two Nodes (starting and ending) and may be a Ring-edge; > Ring is composed of a Sequence of Ring-edges with a > 'succeeding/preceding' relation on them; > A Ring may be used as a Face-Ring, and may be Internal or External > (only one external allowed); > a Face is bounded by a Face-ring, and may itself be used as a > Shell-face: and a Shell is composed of Shell-Faces. Clear; google found http://www.geovista.psu.edu/sites/geocomp99/Gc99/037/gc_037.htm which is consistent with above and talks about the concepts of Node, Edge, Face, Ring, Volume, Shell and the relationships node-edge, node-face, node-volume, edge-face, edge-volume, face-volume It seems straightforward to OWL-ize those... > USIGS didn't get into the 3-d homology stuff, but you can get that > from a topology textbook. Basic idea: treat each edge piece as having > an orientation and say that opposite orientations cancel. Then add up > all the edges of all the rings of all the shell-faces. If the result > is zero, the shell (surface) is closed; if not, then there are holes > in the surface and the surplus edges are the edges of the hole(s). > For an example, orient four triangles clockwise, join their edges > into a tetrahedron so that matching edges cancel. Now remove one of > the four sides and do the edge arithmetic again. I bet there are > algorithms based on this stuff built into the silicon hardware of an > X-box these days. Right and I believe it is possible to write down a set of N3 implications which can be queried to get such kind of evidence... -- Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/
Received on Tuesday, 20 January 2004 16:50:34 UTC