Monday, January 16, 2006
[350] ß parallel discussion in “national debate” bead thread
This is part of a discussion
that will be moved to a Wiki page soon.
Ontological modeling with
category theory and convolutions
Post OWL strategy
[347] ß Judith Rosen’s recent communication
Communication to BioPAX working group;
I agree that OWL mechanisms support this same-as identification and that under some circumstances an OWL query will produce what appears to be an inference of the induction or abduction type. For the record the key is "appears", and in some specific cases this appearance can be reasonably justified.
Allen, I accept your two conversation distinction. However...
The mapping may in fact not only occur in the instance of a mapping between two different information bases (I do not use "database" as I want to reserve "database" as something that can be addressed uses SQL (my preference). There is a second case of a mapping, and that is "self to self" mapping.
I would want the BioPAX information base to be called an ontology mediated information base, or something like that.
One of the properties of the referential information bases (RIBs) is that they work like hash tables with the strings being the hash key, and a set of pointers in the payload of the hash container being the possible "different" semantic indicators. The IDs are then those used in everyday language, with a focus on phrases and non adjacent co-occurrences also being coded as IDs.
see figure 8
http://www.bcngroup.org/area2/KSF/Notation/notation.htm
If all IDs are unique then there is nothing for the hash payload to run reconciliation orchestration on. However, the hash container also has a pointer to the definition of the concept(s) having been associated with the ID.
It turns out that one can treat a string as a base 64 number... 52 regular letters and some fillers. Then these strings themselves are ordered uniquely, and thus we have a keyless hash table (this work was given a 2003 Patent) surprising... but correct in my opinion because the result is a radical reduction in the machinery needed to get all of the functions of a hash table with great less space and with very very fast convolutional processes.
The point is that such an encoding of IDs (having possible multiple contexts) is organizationally free; ie it is schema independent. So it can encode any relational database, or any set of RDF triples (the long W3C URIs create a problem) and recreate the database or RDF repository without loss. But if one is staying with spoken language, then one has a very surprising new technology (that has not been discovered by the mainstream).
Such an organizationally free encoding of n-aries, or RDF triples can be given different interpretations by changing the terminology reconciliation parameters. I know this sounds a lot like pure theory but bare with me a moment.
Biological science suggests that
The "meaning" of an aggregation of some set of reactants depends on
1) internal causes from the reactants - depending on internal state (which might not be measurable)
2) conservation laws
3) external "environmental" "affordances, such as field dynamics or allosteric mechanics.
I have called any categorization engine that takes these into account separately a "tri-level" architecture.
Within such an architecture the BioPAX data ontology with data can be interpreted using variation of parameters, reflecting the known science. These variations themselves are likely to be in correspondence to naturally occurring categories and thus small in number ( < 100).
Here the meaning of a ID will shift, and a translation in the semantic indicators would be achieved using a single data source.
But I certainly agree that for most, this seems a long ways off.
The take home message is that BioPAX done like Alan is thinking about (as well as others here) will result in a data source that can be harvested into a RIB and translated into an ORB and in this state interact with things like the Mark 3 knowledge processor when ever Dr Ballard finished the Mark 3 editor.
There are often times where the ingredients for a break through have been assembled and then suddenly there is something new. This is what we are anticipating.
Anyway back to the details... (having hopefully suggested what the horizon might look like)
Dr Paul Prueitt PhD (mathematics and quantum cognitive neuroscience)
The Taos Research Institute
Taos New Mexico