Monday, February 20, 2006
Generative Methodology Glass Bead Games
ruleML and object oriented ruleML (rule markup language) are both from one information paradigm and descriptive logics are from a different paradigm. Like French and Germany languages, it has become useful to think about re-use of information models developed in UML (for example) and re-express this information as a descriptive logic based class-subclass hierarchy. The reverse is also interesting, as well as translations between information described in these two paradigms and the relational database paradigm of Codd.
forums on the current translations efforts might not really be set up as this is only recently becoming more important. One area of interest comes from the ebXML rule and information modeling efforts.... in the context of web services. This is sometimes called Service Oriented Architecture (SOA).
Service is the key to "what is next". From issues related to environmental management, the New War, preparation and responses to difficulties such as the possible pandemic flu; all need a universal translation mechanism between information standards. Because the value proposition is so high, we see very rapid progress being made (sometimes). Maybe the conditions are right for a quantum leap in capability to model information and provide 1) translatability, 2) secure information channels, and 3) transparency. (See Roadmap from the BCNGroup)
Translations between the XML / object oriented standards and the descriptive logics standards and the relational databases are often called mappings. I have recently seen a foundation for a possible proof that any thing expressed in any one of these paradigms can be completely "translated" into any of the others. But this proof, if one is possible, is heuristic in nature and has to rely on construction... ie general principles and "rules" that can be applied to one standard to produce the "finite state machine" which is the "same" information but is expressed within a different standard. Clearly these constructions are not available at this time. Perhaps in one or two more years?
There is work on something called FERA (Federated Enterprise Reference Model) at OASIS that might be a forerunner of a engine (middle ware) that enables a universal translation between any of a class of standards, thus FINALLY providing "structural interoperability". Behavioral analysis is handled separately and with a different thought process, in FERA and in the emerging theories about the meaning of structurally encoded information - whether object oriented, Codd-type relational database, descriptive logic based ontology, or the formative ontological modeling based on n-aries.
The excitement comes from the possibility that information can be indexed with a flexible and more extensive structure available in object oriented information models with rules, and the functional or behavioral (semantics) approximated by OWL. Topic maps fit in here somewhere, but a TM use model has not been (to this point ) competitive with either
1) object oriented,
2) descriptive logic based subsumption ,
3) relational database management systems with efficient persistence and SQL structured against schema that captures a specific IM or ontology informational index.
This is indeed an exciting time. Theoretical issues arise in the context of closed world assumptions, partial open world assumptions, and formative situational ontological modeling.
Formative situational ontological modeling is fully consistent with an open world assumption that allows on only incomplete and even new structure to data instances to be added into persistence, but allows the "meaning" of data and structure to be altered in real time.
Of course, this formative situational ontological modeling is mostly a theoretical paradigm ... but one which might be the limiting distribution of the current rapid work on information integration and use.
I offer these comments with a realization of how very difficult it is to grasp all that is going on and communication what is grasped correctly. So I may make mistakes now and then.
We thank you for you question.