December 6, 2005
Center of Excellence Proposal
à
[bead thread on curriculum
reform]
The Taos Institute
(on the possibilities)
Communication to Protégé:OWL eforum : protege-owl@smi.stanford.edu
Protege team,
I wish to express my thanks to the Protege team for the work that has been done, up to now.
My comments are reflective of a widely felt frustration over software standards in general, and the absence of the "ontology community's" agreement on foundational theory and on elementary concepts.
RDF starts out with a simple notion, that of the RDF triple < subject, verb, predicate >, and with this there is a direction to an expression of certain types of knowledge. An RDF statement is a sentence in natural language.
Hierarchical graphs are expressible when the verb becomes "is-a". Once this hierarchical graph is established, it is possible to use class and set membership to produce deductions. The addition of properties < something, has the property of, something else > and attributes < something, has the attribute of, something else > and relations < something, is related to, something else >, allows a query language to be developed.
If the description of how an ontology editor works would start out this way, plus explicitly states and helps the learner to realize the limitations to general knowledge representation that is imposed by hierarchical "taxonomies", and then go on and exploit the RDF paradigm in applications such as a class of web services, then we would have a functional and stable RDF paradigm. Intellidimension Inc is a three person company who has done precisely that. But Intellidimension is not mandated by X, where X is a government activity or a business activity depending on the notion that a standard has to be used. I am just using this one small vendor as an example of what I will likely use on my next project; rather than protege.
RDF is a concept, that can be described in a few pages. If its limitations are recognized then the common use of this well defined concept would make a great deal of difference to government and industry processes.
The Semantic Web community will not stop and allow the rest of society to benefit from RDF repositories within imposing the notion that intelligence agents (computer programs) and logical processes matches the task of modeling generic real world processes; such as metabolic processes involved in reactions to flu. Of course there is some benefit, but this benefit is made inaccessible by trying to reach too far with the RDF constructions.
The esoterics, like what is attempted by the SOAR group, or other academic groups should not be grouped in with the core "yes sir, this works and is what we say it is". The SOAR work attempts to characterize cognitive interfaces. A nice paper (2004) from this group is at:
http://acs.ist.psu.edu/papers/haynesCR04.pdf
I noticed the support from the Navy. The military support for this type of work is decades deep. However, this work does not go to the core reason for being, as stated by those who advocate the Semantic Web paradigm; ie data exchange interoperability.
The attachment of the first order logic and the embedding of RDF and this logic into a specific software architecture (Java, or J2EE, etc) is where (I feel) the complication and confusion start.
Again, I wish to express my thanks for the excellent work done up to now; and would like to encourage the team members at Stanford to bring from the current Protege something simple, easy to use and restricted in scope to a core set of capabilities.
This can be done if the team wishes.