W3C home > Mailing lists > Public > www-rdf-interest@w3.org > April 2001

Re: 'Meta' terminology (was RE: N3 contexts vs RDF reification)

From: pat hayes <phayes@ai.uwf.edu>
Date: Fri, 27 Apr 2001 12:02:45 -0500
Message-Id: <v0421011bb70f48072100@[205.160.76.196]>
To: "Chris Fox" <cfox@lds.com>
Cc: www-rdf-interest@w3.org
>I would want to add that:
>
>1) All metadata is in fact data. How often do we explain, "it's data 
>about data"
>?
>2) Meta-metadata is data about metadata, and therefore, falls into 1.
>3) "Meta" normally means beyond, not above.
>4) About and above are not the same concept.
>5) Anything stated in language, even if it is "about" language, remains
>language.
>6) There is no metalanguage (unless you're a Chomskyan).
>7) If there were a metalanguage or metadata, it would be only ever be
>known/understood/respresented as language or data.
>8) Cf. 6, there is no metadata.
>9) Any metadata, if such a thing were possible, would be something like this:
>what the data respresents external to the data structure (e.g., me 
>as opposed to
>the record with the fields pupulated by referents to things about me).
>10) "True" metadata would need to be pointed to, not respresented in a data
>structure.
>11) Metadata is therefore only "meta" from the frame of reference of 
>the data it
>describes.
>12) There is no ontological "meta."
>13 "Meta" is therefore a design decisison.
>
>Many of us know this, but it's good, once in a while, to reawaken to the
>difference between maps and territories, the difference between 
>topics and their
>occurrences.

I really do not know what you are talking about, but much of what you 
say doesnt make sense, or is trivial and beside the point, or is 
simply wrong, if I understand it at all. (I do know the difference 
between maps and territories, having cut my milk teeth on Korzybsky.)

1. Of course metadata is data, just as a metalanguage is a language. 
But the (inappropriate) use of a mass noun does not make all useful 
distinctions vanish. One piece of data may be metadata for another 
piece of data. That is a potentially useful, and certainly often 
meaningful, relation between them.
2. See 1.
3. There is no "normally" in this game. What you consider normal 
usage simply reflects your background reading and interests. It has 
about as much significance as your 'normal' accent: it only tells the 
rest of us where you are coming from. In logic and technical 
philosophy, "meta" has a fairly exact meaning, certainly in the 
context of "metalanguage".
4. I agree entirely.
5. True, but irrelevant: see 1.
6. Plain false. The technical literature abounds with metalanguages. 
(Some languages can act as their own metalanguages, in a sense. LISP 
is the classic example. But this needs care. Strictly speaking, LISP 
and other interpreted datastructure languages are 'reflexive' only in 
a derivative sense, which arises from the fact that LISP code is also 
LISP data. If you try to give a genuine semantics for LISP, as 
opposed to simply an interpreter specification, this reflexivity is 
much harder to describe properly.)
7. See 5.
8. Well, I'm not yet quite sure what 'metadata' means, but I doubt 
that this is true.
9. Completely confused. There are two distinctions which need to be 
sorted out to have a rational discussion here. First, there is the 
distinction between a language and its semantics (how it is 
interpreted: what it represents externally to it.) Then there is a 
distinction between one piece of data/language and another piece of 
data/language which is about the first one, which is the meta- 
distinction. These are not the same distinction, though they are 
related, in that if A is about B, then B (not the content or referent 
of B, but B itself) had better be part of the semantic domain of A. 
(This, by the way, is why I am very leery of the widespread use of 
'reification' in RDF discussions. If this really does mean what it 
seems to say, then reification cannot possibly be used to encode 
negation, disjunction and so on. )
So what you say above is backwards. If A is metadata for B, then B is 
in A's 'external' semantic domain, not the reverse.
10. See 9. "True" metadata would have the data (with respect to which 
it was meta-) in its semantic domain. Which is another way of saying 
that it would be about that data.
11. Too vague to be meaningful. I think I agree, but so what?
12. Just plain false, if it means anything at all.
13. True, but vacuous. All of this stuff is a design decision, since 
we are designing the languages. If you mean, therefore it is 
arbitrary or meaningless, then that is both wrong and dangerous. 
Getting these issues confused is like being careless about critical 
mass when dealing with plutonium. Almost all the known-to-be-fatal 
paradoxes arise from this kind of carelessness.

Pat Hayes

---------------------------------------------------------------------
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 Friday, 27 April 2001 13:02:58 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:49 GMT