Re: I have a trouble with The RDF Model

At Sun, 26 Nov 2000 14:07:14 -0800 Stefan Drecker wrote:
>At 02:15 PM 11/26/2000 -0600, pat hayes wrote:
>>Stefan Decker, replying to Seth Russel, wrote:
>>
>>>a triple is unique - nobody can distinguish
>>>between the triples [Bush, wonThe, Election]
>>>and [Bush, wonThe, Election] per se.
>>
>>That is correct if 'triple' means an abstract mathematical object. 
>>However if these 'triples' are syntactic entities (kinds of 
>>expression) then one most certainly can distinguish two distinct 
>>such entities which have identical structure. This is called the 
>>"type-token distinction" in linguistics: its the commonplace 
>>observation that one can say or write the same 'thing' twice (two 
>>tokens with the same type).
>
>how are they distinguished?

By being at different positions on the page, or being spoken at 
different times or in different circumstances, etc.. There is nothing 
very deep here, its only a technical way to talking about the 
obvious: one can write (or say, or put on a webpage) the 'same' thing 
more than 'once'. Usually the meaning of the expressions of the 
language can be given independently of the particular token; those 
that cannot are often called 'indexical' (eg words in English like 
"now" which have a different meaning every time they are used.) So in 
your original reply, I am not sure whether you are saying that a bare 
triple is like an indexical, or that the 'index' (after the "at") is 
part of the language. In the latter case the language is not 
indexical, in the former case it is; one has to treat the semantics 
rather differently in the two cases.

>>>However, as you have already observed, the source of
>>>the triple might be relevant for believing a fact or not believing a fact.
>>>A model theory (assigning true or false)
>>
>>That isnt quite what is meant by a model theory. (It would be a 
>>very trivial model theory.)
>
>Nevertheless - a mapping from a kind of expression (a mathematical 
>object) to some interpretation.
>
>>>therefore has to include the source of the triple.
>>
>>No, the model theory simply assigns interpretations to the 
>>syntactic constructions; it does not control the syntax. If the 
>>'at' construction is part of the syntax then the model theory 
>>should assign it a meaning, and if it is not part of the syntax 
>>then it should ignore it. So, is 'at' part of RDF syntax or not?
>
>I never claimed that the model theory should control the syntax. I 
>just said that the
>domain of the interpretation mapping has to include the source of 
>the triples into account.

The domain of a model theory, more or less by definition of the term 
"model theory", is the expressions of the language (or perhaps more 
exactly, the parsings of those expressions according to the syntactic 
rules of the language.) So I repeat: are you saying that the 'at' 
assertions are part of RDF, or not?

>>BTW, a model theory should assign a meaning (referent) to every 
>>well-formed expression of the language, so even if the language 
>>contains expressions of the form
>>([Bush, wonThe, Election] at RobustAI)
>>the question still arises of what exactly the subexpression
>>[Bush, wonThe, Election]
>>denotes, since it is not a truthvalue (it would probably be 
>>something like a function from the things denoted by the 
>>expressions after the "at" to truthvalues, ie a predicate on those 
>>things, whatever they are.)
>>>That means it should act on syntactic constructs like
>>>([Bush, wonThe, Election] at RobustAI)
>>>and
>>>([Bush, wonThe, Election] at Electoral College).
>>>These constructs are different even if the triples are identical.
>>>
>>>This violates neither the uniqueness of triples nor
>>>the Law of the Excluded Middle.
>>
>>Yes, but you seem to now have changed the language into something 
>>else, and the original question (and this mailing list) is supposed 
>>to be about RDF. If someone asks whether there is an English word 
>>for "Schadenfreude", it isn't much use to tell him that there is 
>>one in German.
>It was always clear, that RDF-interpretations and applications have 
>to look at the source
>of RDF triples for judging the quality and truth value of statements 
>- thus this belongs to RDF.

Talking of what 'belongs to' something is just muddying water which 
is already so murky as to be hard to see through. Would it be 
possible for you, or someone connected with RDF, to give a precise 
specification of what exactly *is* the syntax of RDF? Until someone 
does, none of this discussion is really worth having, since it is too 
imprecise to make any rational sense.

>The open question is how to represent the source in RDF.
>The result of an earlier discussion was to represent it by reification.
>(See: http://www-db.stanford.edu/~stefan/updates.html for links to 
>the earlier discussion.).
>The language has not changed  -  I used  ([Bush, wonThe, Election] 
>at Electoral College) as an abbreviation for
>{[Bush, wonThe, Election])
> [id1, type, statement]
> [id1, subject, Bush]
> [id1, predicate, wonThe]
> [id1, predicate, Election]
>[id1, at, Electoral College]
>}

OK, that helps, certainly. Now, does *that* have a model theory? In 
particular, what does 'id1' denote? (Let me point out a possible 
problem here. If your answer is going to be that 'id1' denotes the 
triple [Bush, wonThe, Election], then we have a very deep problem in 
giving a normal model theory to RDF. The problem is that according to 
this rule, RDF names may refer to pieces of RDF syntax, so the 
language is highly self-referential. This doesn't necessarily make it 
impossible to give a model thoery for RDF, but it does take it well 
outside what might be called the 'normal' spectrum of logical 
languages, so makes the question of what exactly the semantics is, 
both more urgent (since it isnt what it superficially appears to be, 
ie a simple assertional language) and more difficult. This is why I 
tried to focus your attention on the issue of what exactly the 'bare' 
triple denotes, since it would be a lot easier to give a model theory 
to the language if 'id1' denoted that, whatever it is, rather than 
the triple itself.)

>(although I still would vote to extend the RDF model)
>
>>>One could vote to include the source of the triple into the RDF datamodel
>>>itself.
>>>Former discussion of this question in the rdf-interest group resulted in the
>>>expressed opinion to use reification for this purpose
>>>(see ).
>>
>>I have a slightly more basic trouble with the "RDF Model". I really 
>>cannot understand what it is supposed to be a description OF.  Are 
>>these 'triples' to be considered syntax or interpretation? Since 
>>RDF has an XML-ish syntax which is quite different (it involves 
>>many angle brackets and quotation marks, for example), the 
>>'triples' are apparently not the syntax. So I presume that they are 
>>intended to be part of the interpretation of the syntax, ie the 
>>semantics of RDF is defined in terms of abstract entities called 
>>'triples'.
>>But if that is so, then Stefan's reply to Seth, above, doesn't make 
>>sense, since obviously semantic interpretations aren't the kind of 
>>thing that can be located on websites, or which have 'sources' in 
>>this sense, or which are assigned truthvalues; and Stefan seems to 
>>refer to the model theory *of* the triples, rather than to the 
>>triples as constituting the model theory.
>You seem to assume, that on the web we will have one (or at least a 
>small number)
>of sophisticated language(s) that will be sufficient for everybody
>to express their data/information/knowledge it.

No, I am not assuming anything of the kind. I am talking about one 
language, indeed; viz.. RDF, since this is the topic of this 
discussion. But this discussion about RDF is not about the sociology 
of the Web.

>I don't believe that this will happen. We have a multitude of 
>formats already -
>and we will see even more different languages for various purposes.
>Naturally, this will lead to a multitude of interoperation problems.
>
>The best one can do is to provide a common ground for all the 
>different languages, such that
>establishing interoperation if as cheap as possible.
>(even if the others do not base their languages on top of it - it 
>can be used as a more
> abstract base for technology development).
>
>From database research, it is well known, that semi-structured data
>(a graph form) is useful for mapping between heterogeneous datasources.
>(see eg.
>Papakonstantinou, Y.; Garcia-Molina, H.; Widom, J.
>Object Exchange Across Heterogeneous Information Sources
>1994,ICDE '95
>http://dbpubs.stanford.edu/pub/1994-8
>
>Correspondence and Translation for Heterogeneous Data,
>Serge Abiteboul, Sophie Cluet, Tova Milo, ICDT, 1997.
>ftp://ftp.inria.fr/INRIA/Projects/verso/VersoReport-102.ps.gz
>
>or a newer paper:
>Representing and Transforming Model-Based Information
>by Shawn Bowers, Lois Delcambre
>ECDL 2000 Workshop on the Semantic Web
>21 September 2000, Lisbon Portugal
>http://www.ics.forth.gr/proj/isst/SemWeb/program.html
>
>RDF is exactly this kind of semi-structured data - even if you think the
>syntax is ugly (it's just one particular syntax)

Sure, never mind my 'ugliness' jibe. But more to the point, what (in 
the light of my question) *is* the syntax of RDF, then? I have no 
problem with this syntax being graph-like or diagrammatic or abstract 
or almost anything else, I just want to know what it IS. Until we 
have a clear answer to this, then any talk of semantics or meaning is 
bound to end only in confusion.

BTW, what you call, 'semi-structured data' is a simplification of an 
old idea (c. 1965) due to John McCarthy, which he called 'abstract 
syntax', and which I was implicitly referring to above by saying that 
the model theory is best thought of as applying to the parse trees of 
the expressions of a language rather than to the expressions 
themselves. You can find references on his website.

>So the triples at first can be understood as assertions.
>But the triples are also the syntactic material we are dealing with.

OK, so the triples are the (abstract) syntax; is that correct? That 
would be a very good answer, since it would allow a much improved 
surface syntax to be also called 'RDF'.

>>So I am left completely confused about the meaning of RDF, as 
>>indeed I have been ever since first meeting it. I have been 
>>assuming in the DAML discussions that it is basically simply a 
>>syntactic specification (possibly the ugliest ever devised by any 
>>human being, but let that pass) without any actual semantics. 
>>However, some people seem to think that it has a semantics. Can 
>>anyone point me to a specification of a model theory for RDF?
>
>I would say, given a specific set of RDF triples, most of this apply:
>
>Raymond Reiter. Towards a logical reconstruction of relational 
>database theory. In M. L. Brodie et al. (Hg.), On Conceptual 
>Modeling, S. 191-238. Springer, Berlin, 1983.

That is a general paper about methodology, not a specification of a 
model theory for a particular language (which wasnt invented until 15 
years after the paper was written.)

I gather, then, from your answer, that RDF does not, in fact, yet 
have a model theory (as I have long suspected).

Pat

---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes

Received on Monday, 27 November 2000 10:43:32 UTC