W3C home > Mailing lists > Public > semantic-web@w3.org > December 2008

Re: Ontology for points in a three-dimensional space

From: Toby A Inkster <tai@g5n.co.uk>
Date: Mon, 8 Dec 2008 23:41:48 +0000
Message-Id: <81B9F8AF-4B50-4BF8-A60C-70D7919DADB8@g5n.co.uk>
Cc: Semantic Web <semantic-web@w3.org>
To: Story Henry <henry.story@bblfish.net>, Norman Gray <norman@astro.gla.ac.uk>

On 8 Dec 2008, at 23:11, Story Henry wrote:

> Hehe. Cool, but I don't know if you need to make a new class for  
> three dimensional points. All points are 0 dimensional, see
>
>    http://blogs.sun.com/bblfish/entry/the_10_dimensions_of_reality

Points are by definition 0-dimensional, but their position can be  
recorded in an N-dimensional space. Although I'm only interested in  
recording their position in 3 dimensions, an ontology which allows  
them to be recorded in N-dimensions would be fine too.

> On the other hand yes, I can see that if you are using euclidian  
> geometry you may need specific euclidian x, y, z coordinates. Those  
> differ from the x, y relations of geo coordinate space, which  
> occurs on the surface of a sphere.

The W3C geo vocab includes a "z" - altitude in metres. Essentially,  
the W3C geo vocab records (x,y,z) co-ordinates in one particular non- 
Euclidian geometry - i.e. WGS84. I am after something that does the  
same, but can be used in other, Euclidean and non-Euclidean  
geometries - the ":plane" predicate would be used to state which  
geometry is being used. (The plane may, for example, define where the  
origin is, which direction the axes lie in, the units being used for  
each axis, plus any other pertinent information.)

Norman Gray wrote:

> What's your application?

>

The idea is that I could do things like this:

<#moonbase>
	foaf:based_near
		[ a geom:Point
		  geom:x 56.1 ;
		  geom:y -12.9 ;
		  geom:z 0 ;
		  geom:plane <#lunarPlane> ] .

<#lunarPlane> a geom:ReferencePlane ;
	rdfs:isDefinedBy <http://adsabs.harvard.edu/abs/2002cosp...34E1305W> .

Without having to create separate ontologies for each reference plane.

> How general do you want your coordinate system to be (there's a  
> whole world of positively delicious complication waiting to be  
> explored there!)

I really just want three numbers with an associated frame of  
reference which may be defined by a link or short descriptive text.  
Something like that should be simple and flexible enough to work for  
any three-or-less-dimensional geometry - even exotic things like co- 
ordinates on, below or above the surface of a torus.

-- 
Toby A Inkster
<mailto:mail@tobyinkster.co.uk>
<http://tobyinkster.co.uk>
Received on Monday, 8 December 2008 23:42:25 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 21:45:26 GMT