one line RDF abstract syntax. was RE: RDF core: abstract syntax and semantics - an attempt

Graham,

The problem with the way RDF reification is implemented is just that -- it
is an implementation technique for attaching an rdf:ID to a statement using
a set of triples.

Since a statement is itself a resource (indeed the reification _mechanism_
is but one way of attaching an rdf:ID onto a statement), why not simply
allow a statement to be the subject or object of another statement, e.g.

world ::= collection*
word ::= (URIref|qname|literal|variable)
term ::= (word | statement | collection)
statement ::= [(URIref|qname), term, term]
collection ::= {term*}

by representing a term in this manner, the following substitution can be
made:
term ::= (word | [term,term,term] | {term*})

and then the single line:

term ::= ((URIref|qname|literal|variable) | [term,term,term] | {term*})

and we describe the world as:

world ::= ({((URIref|qname|literal|variable) | [term,term,term] |
{term*})*}*)

one can readily represent patterns for something like KIF sentences using
such terms. The question remains:

what is the purpose of the RDF mechanism of reification beyond an
implementation technique for attaching an rdf:ID to a statement? We should
discuss the merits of RDF reification in such light, i.e. how good an
implementation technique is it?

-Jonathan


>
>
> At 01:52 AM 6/22/01 +0100, Danny Ayers wrote:
> >Hey - good work. There are at least two other attempts to do the
> same that
> >have been by this list, but neither promoted the reification (or is it
> >quoting) as a hook - some more non-normative but informative
> words would be
> >appreciated.
>
> Thank you.
>
> The reason I tackled this myself as yet another attempt at RDF semantics
> was, in part, to try and see how what I view as the simplest possible RDF
> core could work out;  particularly the "reification" bit.
>
> I don't know if what I am calling "reification" is what true logicians
> would call "reification".  But it seems to me that what I have
> modelled is
> quite distinct from "quotation":  at no point are statements in the RDF
> language made part of the domain of interpretation.
>
> If this is felt to be a useful contribution, I hope to engage in ongoing
> discussions and clarifications in response to feedback.
>
> Another of my reasons for preparing this note was to set out what I have
> recently come to understand is the relationship between abstract
> syntax and
> semantics:  I have had some private discussions about syntax and
> semantics
> (e.g. why is "reification" drawn out as a separate syntactic
> construct when
> it can also be expressed as a collection of ordinary triples).  I
> think it
> is reasier to see why this may be useful when looking at the way
> semantics
> are defined in relation to the syntax.  To quote from the Scott/Strachey
> paper that I reference:
> [[[
> In all cases there is a syntactical definition in several clauses.  The
> semantical definition is "syntax directed" in that it follows the same
> order of clauses and transforms each language construct into the intended
> operation on the meanings of the parts.
> ]]]
>
> I'll also note that the use of square brackets in the semantic
> definitions
> to clarify the distinction between things in the semantic domain
> and things
> in the domain of interpretation is a stylistic technique I
> adopted from the
> Scott/Strachey approach.  For me, this helps to keep at the forefront the
> fact that we are dealing with a number of different kinds of thing:  a
> language, exp[ressions in that language, things from a domain of
> interpretation, and a metalanguage to0 describe the domain of
> interpretation.
>
> Beyond this, it is my hope that the "core RDF" that I have targeted is
> sufficiently simple that the expression of semantics are easily
> understood,
> and can be seen as intuitively related to elements of RDF that we
> know and
> love (or hate).
>
> #g
>

Received on Friday, 22 June 2001 08:43:23 UTC