Re: Scaling/size of geographical entities

Thanks Joshua.  With a basic glimpse on this, I would hope that the geo 
ontology is extended with a geo:radius attribute: it is a low hanging 
fruit while having feature types is probably an overkill.  To specify 
the feature type of something, it would be sufficient to make it an 
instance of some class, e.g. subclassing SpatialThing.

However, I wonder how do visual/interactive systems deal with the 
scaling issue right now, if they only have latitude/longitude values? 
Can anyone in this group explain at what zooming level they start, e.g. 
if they show a mash-up page for real-estate development?  A "standard" 
radius attribute could be attached to either classes or instances and 
then used to automatically select an appropriate zooming level.

I can imagine that having no such standard facility for scaling will 
likely lead to drifting solutions and partition the community.  Starting 
small is fine, but could also be a wasted opportunity.

Holger


Joshua Lieberman wrote:
> I would suggest, if you haven't already, to have a look at 
> www.georss.org, which we intend to use as the source of the geo update. 
> There are two attributes defined in this model which could be useful for 
> communicating scale. The first is the featuretypetag, since feature 
> types such as cities may carry an implicit scale with them. The other is 
> the radius attribute, which can define a radius of interest about a 
> point or other geometry. One could go even farther and define an 
> explicit semantic for point with radius (e.g. <is-located-at+aoi> ) 
> using the relationshiptag attribute.
> 
> You also point out an area for future geospatial ontology development 
> activity, that of context or viewpoint. For now, however, it should be 
> practical to start small.
> 
> Cheers,
> 
> Josh
> 
> On Jul 20, 2006, at 5:19 PM, Holger Knublauch wrote:
> 
>>
>> All,
>>
>> long/lat coordinates are very useful to describe points but they don't 
>> tell us anything about the scale/size of the item.  For example, if we 
>> attach a coordinate to a city then this means something really 
>> different from a coordinate on a specific restaurant in that city.
>>
>> The practical problem I am facing right now is that I am developing a 
>> map browser for our ontology editor, so that it can show Google maps 
>> for the selected object (assuming it has coordinates).  However, when 
>> the user navigates to a given resource, how can I determine a useful 
>> zooming of the map - here I would need something like dimension or 
>> area to get an idea of whether to show the whole city, a suburb, a 
>> block, or a street or backyard.  What about introducing an optional 
>> but recommended property for scale or size?  The onion stuff could 
>> help, but I don't see a standard property for this in the vocabulary.  
>> Alternatively there could be an area class, but this may complicate 
>> querying and using the vocabulary.
>>
>> I guess this may return the group to the discussion about the size of 
>> the new vocabulary, and where to set the boundary.  But my feeling is 
>> that some scaling information is critical for any form of visualization.
>>
>> Thanks for any comment,
>> Holger
>>
>>
>

Received on Friday, 21 July 2006 20:27:34 UTC