ORB Visualization
(soon)
3/9/2004 9:20 AM
Dick Ballard makes the following note, and Paul Prueitt’s comments are linked by [*]. (in progress)
Paul:
You are circling close around the right ideas on several of these things.
If you stick to the words you use about a multi-dimensional space, you could be right on target.
But frustratingly you keep dropping back to ordered triples <a r b> and throw away all the dimensionality in pair wise conditional (if ...., then .....) relationships. [*]
Please escape the conditional sentence grammar and go to full space mappings. That would move you into the n-ary theory bundles and constraint browsing. [*]
You are thinking that <a r b> is general, because you can parameterize r(p1, p2, ....pn). That is because your SLIP approach is looking only for pair-wise correlation measures. [*]
You can stick with statistics if you want by going to factor or sensitivity analyses and eigenvectors that unlock an unconstrained wider circle of situational and intentional dimensions. [#]
In the declarative n-dimensional space you can exchange independent and dependent roles and reason in any order or direction, e.g. p2 as a function of (a,b,p1,p3,..pn). [^]
You are in the company of physicists who think routinely in all "degrees of freedom". [^]
Dick