January 1, 2006
Additional reading:
Cory Casanave's paper on Data
Access
work on ontology for biological signal pathways
EU’s program to model complexity using ontology
Azamat,
I read your short EIS standards for ontology pdf a second time. The pdf is attached again for review by the cc and bcc list. (Link)
First, as I read, I felt often that I would like to remove the discussion about motivation and clarify/extend the discussion about what the four principles are for, and why this creates a framework. This is so that those who already believe that "something, other than what we have" is possible might just focus on the framework aspects.
I can also imagine extending the reduced text by using
1) some elements of the BCNGroup Roadmap (hash table methods for data encoding, formative ontology as a consequence of human-computer interaction, organization of an ontology module hub)
2) some elements that were in a 2003 DARPA proposal I wrote fro SAIC.
The hash table encoding should be open source and not encumbered by any proprietary property.
Our compensation would come initially from the EU funding, and then from some type of not for profit corporation that we would set up. The knowledge sharing core concept comes to mind, which you have reviewed already.
In the DARPA proposal we identified a group of 20 leading natural scientists who would have formed a peer review and advisory board for a three year project to develop ontology mediated human discussion between intelligence analysts (The "Glass Box"). The modification would use the same social structure but with a group of natural scientists. We would have members of evolutionary psychology, linguistics, social biology, formal logicians (process calculi and related), ....
We believe that we can get Gerald Edelman to be on board, as well as Paul Werbos and a few other well known high profile scientists. Each of these scientists would be committed to the pragmatics of
1) close world assumption in the development of structural information regarding things that are stable, and
2) the open world assumption about every thing else.
The maturity of having a place for both closed world and open world assumptions is important - and should be part of a curriculum.
The knowledge engineering creates a stable infrastructure - but the architecture assumes that the flexibility implicit in open world assumptions have to shape the user interface.
We feel that knowledge science curriculum development as part of the EU program would be welcomed by the review boards, if the curriculum was part of the usability aspects of our proposals for modeling complexity with ontological models. We have 5 weeks to put the proposal together.
The key has to be "education". To be educators we have to understand something that others need to learn about. We suggest that this "something" has a lot to do with the Rosen definition of complexity, and the misuse of the term "complexity" by the computer science community.
Language use would deconstruct phrases like "artificial intelligence:, "expressive logics", "formal semantics"; and show what is suggested, by AI for example, is not what can be provided, by AI. So the confusion that is in the AI and knowledge engineering communities is shown to be a result of their improper use of natural language. Getting the language right in discussions and in proposals leads to our ability to create a "correct" curriculum where the notion of "machine reasoning" is replaced with a longer description of structural alignments of data... something like "query and retrieval".
"Query" is semantics bound to "asking a question", and this binding is to the everyday experience of humans. This overloading of terms is huge, and without creating a new branding language for what machines actually do, one cannot address the real differences between computer programs "understanding" and human's "understanding".
Robert Rosen's definition of simple systems and complex systems would form a type of ultimate authority on the nuances of the curricular language.
This seems to be a necessary compromise to the knowledge engineers who do not acknowledge any need for questioning whether the closed world assumption is not universal.
I am reviewing the work of a group bioinformatics researchers (BioPAX) who started a few months ago to develop OWL ontology useful for assisting scientific communication about data being repositories in databases. There are about 30 individuals in the working group, and the group dynamics is very similar to the ONTC group in that closed world assumption is considered essential to the work.
I am trying to find some subgroup that is aware of the potential benefit in having an open world assumption over at least some of the ontological modeling.
To achieve clarity on the complexity issue within five weeks we need to start work now.