- From: Sergey Melnik <melnik@db.stanford.edu>
- Date: Sun, 17 Jun 2001 15:42:17 -0700
- To: jos.deroo.jd@belgium.agfa.com
- CC: aswartz@upclink.com, Jan.Grant@bristol.ac.uk, w3c-rdfcore-wg@w3.org
Jos, by saying that logical formulae are out of scope I mean that it is not our job to define a specific logical language on top of the "core". On the other hand, I'd like to see an extensibility mechanism in place that allows doing such things. I think "anonymous resources" as candidates for some sort of logical variables are so limited that their usefulness / specification effort ratio is too low to keep them in the spec. However, I believe that some logical foundation (like model-theoretic semantics) behind RDF may help to avoid modeling errors. For example, once we have some sort of semantics behind the "core" in place, X is-a Variable X hasColor Gray would probably mean that X is a gray variable, which does not make much sense. I wonder, whether we can set up a path that will support giving logical meaning to constructs like EXP is-a LogicalExpression EXP hasSubj X EXP hasPred hasColor EXP hasObj Gray X is-a ForAllVariable EXP mustBeSatisfiedIn CTX CTX contains (Whale_001 hasColor Gray) CTX contains (Hair_001 hasColor Gray) which is still consistent with the semantics of the "core". In the above example, the interpretation of EXP in the domain of discourse is a "logical expression", and X maps to a "variable". Using additional interpretation functions, it may be possible to answer questions like whether the interpretation of CTX (whatever that is) is say an empty set or not. I was just speculating using the above example. I'm sure people like you and Pat Hayes have a much better understanding of the issues involved, and can help to strike a path for defining logic on top of RDF. The specification of a concrete logic is IMO too monumental a task to address it adequately within the RDFCore WG. I'm sorry if what I said earlier sounded like logic is unimportant... Sergey jos.deroo.jd@belgium.agfa.com wrote: > > Hi Sergey, > > [...] > > I don't agree that anonymous nodes should be part of the abstract > > syntax, and would suggest to consider this issue when cleaning up the > > model. I'm convinced that (sufficiently) uniquely generated resources > > serve the purpose better than "anonymous" resources (for instance, in > > the example above you really might want to know whether Person1 and > > Person2 are referring to the same unknown gray thing). Explicit > > existentially qualified variables are IMO out of scope of our work. > > Maybe needless to say again that I have > another IMO, but as our DanC says: > "He Who Does The Work Makes The Rules" > (and you do work a lot, wittness your > GINF work which we apply with-IN AGFA > as AGINFA or A-GINF-A or AG-IN-FA) > and per fundamental human right > "Anyone Has The Right To Fail" > i would be more than happy to fail and > have a better alternative, but saying > that things are out of scope is not > quite convincing. > > I think about a core as being round, > so what's up/down? (more like in/out) > I think there is at least a part > of logic IN the core (maybe EC logic). > I also agree that declarative > programming is indeed a big step, > but in the last 100 years we can > find incredible contributions via the > work of Gottlob Frege, J.A. Robinson, > Pat Hayes just to name a few. > > So I wait for your arguments. > > -- > Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/
Received on Sunday, 17 June 2001 18:16:16 UTC