if I'm understanding the description of this file spec
then, in order to make an application capable of performing searches for overlapping areas in multiple objects,
the data would require a matrix something along these lines:

int model(x, y, z, w)
int model2(x, y, z, w)
int model200(x, y, z, w)
the actual number of models would be a variable in and of itself.
( naturally, that would be a create model function that would check for the total number of models in existance,
prior to creating it )

and each instance of model is an instance of a class, as there is more data than just geometry for it.

the class would contain:
geometry data
texture data
transformation data ( joints for animations )
world positioning data ( yet another x,y,z co-ord set )

my wife uses several 3d graphics / modelling applications, I've noticed that one area where they all tend to be weak is in "boolean" operations.

( union, difference, and intersection. )

the union is the most frequently used, yet the applications fail with the message:
"objects do not intersect"
roughly 75% of the time
I've tested this with object that are 1/3 overlapped in all axis and gotten that message.

so I"m working on a lib to improve these operations.
( and my relationship )

any words of wisdom on my basic premise for the class and matrix structure?

okay, the int is just for this example, float model(x,y,x,w) is more accurate.