Hassoun heeft geschreven in bericht <373CDAD3.6BEABE13@sympatico.ca>...
> ....
>pov objects are created using descriptions of their shapes, position,
>rotation and scale in space, and their surface properties. these
>descriptions can be parametrized (if i can use the verb) in order to
>make the object generic, and relate it to others thus have knowledge in
>them:
>
>e.g. the diameter of a greek doric column is equal to the tenth (1/10)
>of its height.
>
>......
>
>a povray file contains specifications of architectural objects, and
>knowledge extraction (the approach i hope to develop) is to find a way
>to extract all the knowledge (principles, concepts and rules) in this
>file and synthesize it to produce clear definitions of the objects
>contained within. (maybe using C or C++ algorithms, or LISP, or any
>other suitable programming language)


This is interesting, though I have the feeling the POV scene description is very
suitable for this. You need a very strikt way of writing your scene to be able
to extract. And for writing a good scene, you allready need to have (parts of)
the knowledge you are looking for, as you illustrated in your own example.

A different approach could be using XML, write your own Architectural Markup
Language. With this write documents that contain the structure, dimensions and
positions of your architectural objects (kindoff making a database). From these
documents knowledge can easy be extracted. You could even write a conversion
tool that makes a POV scene based on your document. There is lots of information
and free software (mostly Java, Perl) for XML on the net.
http://wdvl.com/Authoring/Languages/XML/
http://www.alphaworks.ibm.com/Home/

I'm intersted in the results you get,

ingo

--
Met dank aan de muze met het glazen oog.

Post a reply to this message