Thursday, February 16, 2006
Generative Methodology Glass Bead Games
n-articulated ontological framework
On using RDF to model biological expression
Link back to part of the “solution” to translations between
RDF / OWL and Models of Information
à- [167]
Concurrent Anticipatory web discussion à [30]
Discussion on expressiveness
Regarding having the namespace reflect the BioPAX version.
Is the issue here that a given version needs to be a fixed "finite state machine", in the sense that if a particular outside program, like an "abstract reasoner" interacts with the version, it should always have the same behavior? This is important because other programs may be designed that have no mechanism for reacting to changes and will fail to react as anticipated. Creating a new name space each time a new version is published then seems reasonable.
Why versions?
If the behavior is not, in some way, as we wish it to be; then a new version makes the modification.
However, "behavioral" analysis of the response of an OWL ontology may be subject to interpretation, and so an earlier version might be precisely what is wanted and needed by a "system" that makes a judgment (based on what ever human reasoning).
One would think, based on the notion of self-referentiality (the standard is expressed within the constraints of the standard) that each version should "know the difference between itself and previous versions". But this information would have to be encoded and that perhaps would be an effort in itself.
Here is where the issue of behavior becomes critically entangled with the notion of "interpretation" or even "discovery".
Andrea and several others are in an excellent discussion about tractability and decidability in the Protege owl forum. If one does not have decidability then one can not complete a behavioral analysis of the reaction of a specific program to a specific version. Right? The same is true for tractability, but for different technical reasons. Right? (I am just trying to get these things settled in my mind).
At
http://www.bcngroup.org/beadgames/generativeMethodology/179.htm
there is some excellent work on a "federation" reference model, that would be useful in conducting "behavioral" analysis of fixed versions in the case where the OWL is already both tractable and decidable.
Are the BioPAX versions both traceable and decidable? (I am asking a question that I do not fully understand.)
The FERA may become seen to have a distinction between structural complexity analysis and behavioral complexity analysis. We are thinking about this distinction in the context of discussions in the Rosen forum.
The contribution of "federation" seems part of the next big development in web services and service oriented architecture.
Sorry for me to go on so long, but I want to say something additional about the task I have taken on related to "mapping" between the SOA-IM (service oriented architecture Information Model) and OWL. This mapping function is also been discussed here between relational database schemata and one of more of the BioPAX versions expressed in OWL.
The federated services to be offered by FERA (and or other proposed or future standards) depends on both
structural complexity analysis and behavioral complexity analysis !!
It seems to me that the ontology community is focused on “structural complexity analysis”, but not on behavioral complexity analysis. In fact the ontology community is focused on a simple finite state machine model (the "ontology") and thus is focused on structural analysis (of the ontology and on the ontologies responses to various programs).
The biological community is (sometimes) attempting to focus solely on behavioral complexity analysis, as when we think of a signal pathway as being a "black box".
If the biological community is also focusing on structural complexity analysis, then we must introduce the structure to function degeneracy analysis that is well known in the biology community be is avoided by the ontology community (because they can not separate the issues related to "tractability and decidability).
Degeneracy is a many to many mapping that is not deterministic and which is observed in nature. Such behavior by a computer program would produce a Turing type halting condition. Yes?
I will say, again to be as clear as possible in the smallest number of words, that degeneracy is seen an a multiple articulated potential that becomes fully constrained as a specific interpretation (mono-articulated) as part of the expression where the expression has emergent properties. This does not mean that finite state machines cannot nicely model stable system behavior. It only means that if we are modeling “arbitrary” reality then we need to “step away from the formalism” now and then. When we, as human’s, step away we can induct a new “simple” machine to serve as a model. (This is “Second School”.)
We are looking for some means to create n-articulated ontology so that this natural phenomenon has a parallel in the computational sciences.