Re: Semantic Layers (Was Interpretation of RDF reification)

Quick notes:

    1) There is, as far as I can tell, no good theories of pragmatics
that are capable of being formalized. "Pragmatics", at least in
linguistics where I come from, is usually a sort of fuzzy "hand-waving"
solution to any hard problem, much as the terms "world-knowledge" and
"common-sense" knowledge are. Whenever I hear the word pragmatics I want
to reach for my axe :) Now, if someone has a good theory formal
pragmatics, go for it and forward me the links off-list, but I  bet is
that pragmatics are too context-dependent to formalize properly. So,
blaming the SemWeb for not doing pragmatics is a bit unfair. 

    2) Re syntax, XML people *like* XML, just in the same manner as
McCarthy was somewhat shocked to find out people actually liked
programming in S-Expressions. Also, RDF is not a programming language.
Therefore, one can do great RDF processing in LISP, where one can car
and cdr (head and tail) to one's delight. It seems like because it has a
formal semantics, once could define a multiplicity of syntax. Lastly,
the SemWeb is a stack, and the logic programming parts of it are in
development right now - I'm sure Sandro can tell us more, and they just
e-mailed their requirements to this list.


John F. Sowa wrote:

> Leonid and Henry,
>
> I am very much in sympathy with Leonid's comments,
> *especially* with the point that pragmatics comes first
> in any kind of design, semantics comes next, and syntax
> should be tailored to the semantics and the pragmatics.
>
>> Let me add some more items to the things you like .
>> - To extend the Triples approach by the Classificaton Theory
>> integrated with Measurment Theory ;
>> - To use the Pragmatics above the Semantics ;
>> - To use the Pragmatism "relativity" between Ontology and
>> Epistemology(Cognitology);
>> - To add the VSM of Stafford Beer
>> (http://www.ototsky.mgn.ru/it/beer_vsm.html)
>> to Upper Ontology ;
>> - To add the "metasystem transition" as "natural generalization" way.
>> ---------------------------
>> See some more details in - http://ototsky.mgn.ru/it/21abreast.htm
>
>
> The details of the systems Leonid cites are very interesting
> and should be considered, but there are also a lot of other
> interesting systems that should be considered.  I don't want
> to obscure the important principle above by quibbling over
> the other details.
>
> The following comment by Henry illustrates the difficulty:
>
> > OWL and RDF are not a syntax. They are a semantics. If you want
> > to make your life easier just use Turtle, or if you want more
> > power than what has yet been standardized, use N3. Here is an
> > atom feed rewritten using N3 (turtle) syntax with the AtomOwl
> > ontology [2]
>
> In politics, statements like the first two sentences above are
> technically known as "spin".  Although I was not involved in the
> design of RDF and OWL, I have read and heard enough about the
> process to know that the designers went through enormously
> complex contortions to shoehorn the finally agreed semantics
> into the syntax policy that was edicted a priori.
>
> Any syntax that is so difficult to use that it requires all the
> effort that Tim Bray and others expended on it just to express
> triples (and which Tim admits was a mistake) has something
> seriously wrong with it.  And if it requires tools as powerful
> as Turtle and N3 to make it humanly usable, that is further
> evidence that something is very seriously wrong.  And if the
> result is so bloated that it requires compression algorithms
> to reduce storage space and transmission time, that is
> abundant evidence that something is horribly wrong.
>
> The *only* argument that makes the slightest amount of sense
> is that the RDF and OWL syntax enables the reuse of parsers
> designed for XML.   That argument is undermined by the very
> simple observation that the notation (R arg1 arg2 ... argN)
> can be parsed with two functions built into many languages.
>
> In many commonly used languages, the two functions are called
> Head and Tail.  But whatever they are called, they provide
> a two-operator parser that comes for free with the language:
>
>  1. For any tuple list L, Head(L) is the first tuple, and
>     Tail(L) is the remaining list.
>
>  2. For any tuple T, Head(T) is the relation name, and
>     Tail(T) is the list of arguments.
>
> Henry's example (copy below) illustrates the kind of metalevel
> notation that is valuable for large documents (and in some cases
> even for a single tuple).  But in many applications, it is
> counterproductive for the metalevel to overwhelm the content.
> If you have a billion tuples (as many systems do), you don't
> want to annotate every last one of them individually.
>
> Recommendation:  Allow RDF and OWL to remain as they are, but
> provide an optional form along the lines of the tuple list I
> suggested in my previous note.  If you want to annotate a single
> tuple in the form Henry suggested, then use the following notation.
> But if you have a billion tuples (or even just a dozen), factor
> out the metanotation and put the content in a simple tuple list.
>
> As Tim Bray commented about the mistakes in the design of RDF:
>
>    It's the syntax, stupid!
>
> John Sowa
> __________________________________________________________________
>
> [ a :Feed, :Version;
>
>    :title [ :value "Example Feed";
>             :type "text/plain" ];
>    :link  [ :href <http://example.org/>;
>             :rel iana:alternate ];
>    :updated "2003-12-13T18:30:02Z"^^xsd:dateTime;
>    :author [ :name "John Doe" ];
>    :id <urn:uuid:60a76c80-d399-11d9-b93C-0003939e0af6>;
>
>    :entry [ a :Entry, :Version;
>             :title [ :value "Atom-Powered Robots Run Amok";
>                      :type "text/plain" ];
>             :link [  :href <http://example.org/2003/12/13/atom03>;
>                      :rel iana:alternate ];
>             :id <urn:uuid:1225c695-cfb8-4ebb-aaaa-80da344efa6a>;
>             :updated "2003-12-13T18:30:02Z"^^xsd:dateTime;
>             :summary [  :value "some text";
>                         :type "text/plain" ]
>           ];
> ] .
>

Received on Wednesday, 29 March 2006 16:47:48 UTC