On Feb 2, 2007, at 10:24 AM, Xiaoshu Wang wrote:
I am very doubtful about the practicality of such an ontology. First, consider the size. RDF is all about URI. If an RDF document has n statement, there will be 3n URIs. If k statements are needed to describe the resolution of one URI, the solution unnecessarily increased the KB k-fold. Image the load for an RDF engine, the impact will be even bigger. In theory, this seems O.K. but in practice, I have serious doubt.
Remember that we are using OWL, which has inheritence. Using an owl hasValue restriction we can have a set of triples specified once, but with the reasoner have be as if each instance has those triples[1]. BTW, I will anticipate the complaint about the OWL reasoner's heaviness. Although we are prototyping with OWL, it may very well be that the constructs we need for this task form a much simpler logic that is tractable[2], or even fairly trivial to work with. For instance, it may be that the run time "reasoner" only has to deal with following subclass, subproperty, and propagating the consequences of hasValue restrictions, and a full fledged reasoner is only used to validate the resolution ontology when it is changed.
In terms of URI's stability, I think what is important will always be taken care of. And so what is broken is not important and just let it be. As for our brain, forgetting isn't necessarily a bad thing. Broken URIs will not necessarily a bad thing for the health of web either. For me, it is more appropriate to discuss the problem in a social, but not technical, context.
A good solution has both components. One of the things that RDF/OWL gives us is a way to encode what the results of our social conventions are. I couldn't agree more that at the basis we need to have a shared understanding of what the problems are, and about the various policies people have when dealing with it. I would argue that a mechanism such as the one we propose will allow for *more* social conventions, because we are not locked into a model where only the W3C defines protocols. And once those policies are agreed upon, encoding the results in RDF/OWL gives us two benefits: First, we are forced to be very explicit about what our decisions are, and often that has the result of discovering problems. Solving the resultant problems can lead to an even better solution. Second, when the end result of our agreement is encoded in a representation language, it is immediately available to be used by automated systems.
Regards, Alan [1] http://www.w3.org/TR/owl-ref/#hasValue-def [2] http://owl1_1.cs.manchester.ac.uk/tractable.html
