Tuesday, December 13, 2005
Proposal: To identify a common, widely
used, and available concept ontology to be used with UDEF (Universal Date Element Framework) and other
finite coding systems underneath.
BCNGroup December 12, 2005
Adam,
I am conjecturing that UDEF (Universal Data Element Framework), and other data coding systems have the nature of a controlled vocabulary.
The commercial world could have dozens of these coding systems and this would be ok. Each one is merely a specific and precise finite state machine. So it is like a data compression dictionary. Having 20 or 30 data compression dictionaries is perfectly ok if one can recognize which one is in use and if one has access that the set of "controlled vocabularies".
It takes a few bytes to say what one uses, and then the transaction partner uses that very small table. The meaning that this data can have can be addressed separately. What we do not have now is complete and consistent data interoperability, and this is what the world needs to have.
If the "formal semantics" is not confusing everyone then a selection of the best coding system in each business domain WILL OCCUR.
We respectfully suggest that this selection is NOT occurring now because of the confusion created by adding logic and formal semantics.
The non-selection is encouraged by natural and normal competition between proprietary coding systems. This non-selection has been going on so long that people, politicians in particular, are taking notice to my argument and arguments like mine.
We, respectfully, suggest that the competitive markets will not, by themselves, solve the global problem of non-interoperability of data codes. One has to have a Nash solution (as in the movie, “The Beautiful Mind”), a solution that addresses common value - where this common value is not seen to be in the local best interest of powerful entities like the W3C. This is why OASIS and IEEE are very important in the current standards struggles.
John Sowa's recent, I think it is a recent insight for him, insight is that we need to find where the controversy is not, and standardize on what is not controversial.
John calls this a Unified Framework (UF). I am suggesting that the UF cannot have any formal semantics, and thus that it must be a coding system like UDEF. This is what a number of non-standard systems have developed, and these are being used in applications where the confusion seems to have gone away.
Semantics is thus left to an interpretant in the present moment. This is a late binding semantics that is not formal because it involves pragmatics (what is available). If the interpretant is a computer program, the program can be designed to test to see if the data is as one would expect data of that type to be. If not, there is a movement within finite state machines until there is a proper fitting. Nothing is "understood". It just works.
The problem with the semantic web, is the same problem that we have all finally understood about AI; it simply pretends that the problem of semantics can be solved with theorem provers. This is too extreme, and yet we continue to allow some individuals (and tenured computer scientists) to hold progress captive. The polemic comes when these individuals say that what the alternative wants is to have no computation. “Negotiations” between finite state machines to resolve the coding system that is being used is computation. Service Oriented Architectures are looking for this type of computation. But we humans should not pretend that this is a machine understanding another machine. The Semantic Execution Environments are being developed to allow human/machine interaction so that question of meaning and practicality are resolved separately and independently from the data interoperability problem.
This ability to not have semantics is an important aspect of how we see coding systems, such as UDEF and (an example that we know) the world customs' Harmonized Tariff Schedule. If there are other coding systems one only needs a look up table to make a translation. It was as if in machine translation there is no Whorf hypothesis (ie no untranslatable terms).
The BCNGroup conjecture is that computer science needs a strictly structural mechanism for providing data interoperability. This mechanism should not have "semantics", certainly not "formal semantics".
Structural interoperability can be defined so that the mechanisms do not have formal semantics.
Like the dollar coin, it is just a coin. It may have a limited type of semantics, as a coin. But my ability to use the coin in a situation has pragmatic aspects that are outside of the limited semantics given to "money".
I discuss the literature a little in this chapter:
http://www.bcngroup.org/area3/pprueitt/kmbook/Chapter2.htm
at
http://www.bcngroup.org/area3/pprueitt/kmbook/Preface.htm
I make a presentation (of this conjecture) more fully. I do apologize that my work is hard to comprehend. I am working to simplify the language. I also apologize if my manor seems authoritarian or arrogant, I do not feel this way inside my heart. I am just trying to be clear about something that has been deeply confused by the incorrect notion that “formal semantics” should be something real. One cannot formalize what things mean, like my love for my family.
We use the term pragmatics to refer to the situated-ness of something. Meaning is thus about categories, and categories "re-occur" so that anticipation can occur. Bob Shaw (Univ of Conn, one of the founders of the ecological psychology research community) talks about quantum theory requiring anticipation to collapse waves into locality. The concepts underlying the ecological psychology (sometimes called ecological physics) community’s definition of anticipatory systems are grounded in physics as well as in the examination of action-perception cycles in animals.
So the use of the terms syntax, semantics and pragmatics each have a separate grounding in a theory of being (ie ontology).
I understand your feelings about OWL and KIF. Somehow the failure to make progress in the data interoperability mess has to be explained. I feel that part of the explanation that we are deriving now includes an understanding that the semantics and syntax must be completely separated so that data interoperability can occur without asking ultimate questions of personal meaning.
This is the proposal.
In the biological realm, most of the key open questions revolve about the structure / function issue, particularly regarding things created (interpreted).
Paul Stephen Prueitt
Taos Institute