[vos-d] abstract/logical semantic metadata

[Reed Hedges]

Karsten Otto wrote:
> However, VOS is not a syntactic file format, but an object system.  
> The object code already represents a form of semantics! Also, there  
> is some overlap in semantic and geometric information, particularly  
> object positions and bounds. So why keep them separate?

In VOS, it's ideal if two different types cooperate and share the same 
child or message with the same name and the same semantics.  But even if 
  they don't have the same name, it's easy to link in an object twice 
with different names.  In the case of properties, the "semantics" of 
using a property is pretty restricted, so sharing properties between 
types is easy.  Right now I'm also trying to make a little toolbox of 
filters for transforming property data too.

So if you have three different aspects to something, say, "3D graphics", 
"physics", and "encyclopedia".  3D properties include position vector, 
size vector, orientation vector. For physics let's just have mass. For 
encyclopedia you have some stuff that tries to define the object in 
terms of other objects that an agent or other system already knows 
about.  If encyclopedia also wants to include mass, but doesn't know 
about the physics aspect, you have two properties:

physics:mass
encyclopedia:mass

These two can really be 2 names for the same property object.

If physics uses grams, and encyclopedia uses kg, then encyclopedia:mass 
property is really just a listener to physics:mass that does the 
conversion, and/or vice-versa.

Is all this useful Karsten to the kinds of things you would like VOS to 
be able to do?



Reed