- From: Paul Tyson <phtyson@sbcglobal.net>
- Date: Sun, 05 Oct 2014 15:27:28 -0500
- To: Michael Brunnbauer <brunni@netestate.de>
- Cc: "semantic-web@w3.org" <semantic-web@w3.org>
On Sun, 2014-10-05 at 20:35 +0200, Michael Brunnbauer wrote: > Hello Paul, > > > > Can you express this argument with triples? > [...] > >RDF(S) alone is unsuited for this, by design. It lacks negation and > >quantification. > > It has existential quantification but lacks negation and a scope mechanism > for both to be full first order logic. > > > Furthermore, traditional logic takes a different approach to argument from > > the the mathematical logic notion of proof. > > I do not know what you mean with "traditional logic" and take it to mean > "forms of cognition exhibited by humans". > Not at all. I mean the instruments of knowledge known, developed, and employed by thinkers (in the Western tradition) at least since Aristotle: in classical education, the subjects traditionally covered in the trivium, particularly logic. See for an introduction Joseph, "The Trivium: the liberal arts of logic, grammar, and rhetoric". > > It should be possible to build for this purpose an RDF vocabulary and > > conventions for use (in the manner of SKOS and OWL). > > OWL is more than a RDF vocabulary. It is a semantic extension of RDF. > With a normal RDF/RDFS/OWL vocabulary, you are stuck with the semantics and > entailment regime of RDF/RDFS/OWL. > > You can escape that without defining a semantic extension by adding additional > semantics via the comments of your vocabulary terms and use a rule engine > instead of or in addition to a reasoner to enforce it. > > But if it would be easy to > > 1) Define a formal system reproducing forms of cognition exhibited by humans > > 2) Define a universal vocabulary (to rule them all) > > Artificial Intelligence would not be where it is now. In fact, many researchers > have adopted the stance that absolute word senses do not make sense (e.G. [1]). > I tend to agree with them. I am not suggesting artificial anything. I am suggesting machine-assisted knowledge transmission. The author notates his concepts, propositions, and arguments in some standard notation. The reader uses a knowledge browser that not only reads the author's logical notations, but allows merging and adding of further propositions to affirm or deny the author's conclusions. > > IMO, any system of the form you are describing can never be universal and > will only be useful for some narrow task. It also seems that what machines can > do with natural language is quickly catching up with what machines can do with > formal languages. When machines can process (for example) the exercises in Joseph's book given to them in natural language, then of course we will not need any special markup or notation. Until then, they need some assistance. (Of course, if they could understand from natural language input what is a title and what is paragraph, what is a reference and what is a foreign word, etc., we wouldn't need XML or HTML either. Stupid machines!) Regards, --Paul
Received on Sunday, 5 October 2014 20:30:15 UTC