- From: Graham Klyne <GK@ninebynine.org>
- Date: Tue, 25 Feb 2003 13:03:13 +0000
- To: Bijan Parsia <bparsia@isis.unc.edu>, Brian McBride <bwm@hplb.hpl.hp.com>
- Cc: www-rdf-comments@w3.org
Bijan, Thank you for your comments. I think your preference here is clear enough, without paraphrasing: [[ >Section 4 of Concepts and Abstract should be struck. ]] This issue has already been raised, and is subject to some debate, and, without prejudicing the outcome of our debate, I'm pretty sure there will be some big changes. Brian: do we need to raise a separate issue for this, or can we tag it onto an existing one? (Your call.) #g -- At 12:02 AM 2/22/03 -0500, Bijan Parsia wrote: >Section 4 of Concepts and Abstract should be struck. > >I have a general argument for this, as well as some specific ones. > >The general one is that specifying the interaction between "social" and >formal meaning, heck, just specifying much of anything about social >meaning is an INCREDIBLY hard task. So either this section is vacuous >(i.e., it doesn't really specify anything and thus can be ignored) or it's >dangerously underthought and underspecified. > >This is aside from the fact that the W3C is probably, by ordinary social >conventions, the wrong authority to be specifying "social" meaning. For >example: > """Human publishers of RDF content commit themselves to the > mechanically-inferred social obligations.""" > >Does this mean, "Hey there, by the laws of the US and most countries of >the world, if you publish RDF on your website, then you're committed to >the mechanically-inferred social obligations"? If so, it's merely >informative (and, in fact, probably false). If it's intended to *make it >so*, then where do you get this authority? > >Goodness, have any lawyers looked at this? > >I mean, the fact that you have written "mechanically-inferred" is *itself* >scarily sloppy. After all, that's reasonably interpreted as *any inference >drawn by a program*, regardless of whether its using RDF(S) entailment or not! > >------- >Detailed comments: > >***Section 4: >"""There are two aspects to the meaning of an RDF graph. There is the >formal meaning as determined by the RDF semantics [RDF-SEMANTICS]. This >determines, with mathematical precision, the conclusions that can >logically be drawn from an RDF graph.""" > >Do you mean that only RDF(S) entailments can be logically drawn from an >RDF graph? That's wrong, as with an OWL reasoner I can draw more >conclusions. And, if we interpret logically strongly, e.g., as >"reasonable" (distinct from formal) then I can logically draw *all sorts* >of conclusions from an RDF graph that aren't sanctioned by the Semantics. > >"""There is also the social meaning of the graph. It is the social meaning >that affects what it means to people and how it interacts with human >social institutions such as our systems of law.""" > >To the exclusion of the formal meaning? Yet 4.4 explicitly connects them. >Indeed, if correct, 4.4 establishes a certain social meaning of an RDF >graph (which includes the social meaning of the formal implications of the >*formal* meaning of the RDF graph). > >***Section 4.1: >"""RDF/XML expressions, i.e. encodings of RDF graphs, can be used to make >claims or assertions about the 'real' world. Such expressions are said to >be asserted.""" > >I'm scared that you use scare quotes for 'real'. I just don't know how to >interpret that. Further more, this section doesn't say HOW one asserts >with RDF expression. Also, it's reasonable to say that only RDF/XML >expression can be used to make claims or assertions. But presumably the >*graphs* are the primary carriers of meaning? > >The second paragraph makes more claims about unasserted graphs with no >specification of when graphs are asserted and *how one fixes the meaning >of the asserted graph*. I mean *fix* it, to a *specific* grounded >interepretation. > >***Section 4.2: > >"""When an RDF graph is asserted in the Web, its publisher is saying >something about their view of the world. """ > >Again, how do I assert a graph (or *not* assert) a graph "in" the Web? >Also, in general if *I* assert something, I'm typically saying something >about *the world*, not my view of it. Also, a publisher may not be, and >may *typically* not be the asserter. Indeed, what is it to be the >*publisher*, in this context? The ISP? The web hosting company? The author >of the document? > >"""Such an assertion should be understood to carry the same social import >and responsibilities as an assertion in any other format. """ > >Well, with HTML, I'm not committed to the mechanically-inferred >conclusions of the HTML dom tree. So....? > >2.3 >"RDF has a formal semantics which provides a dependable basis for >reasoning about the meaning of an RDF expression. In particular, it >supports rigorously defined notions of entailment which provide a basis >for defining reliable rules of inference in RDF data." > >Hmm. A formal semantics does make it easier to reason about the *formal* >meaning of an RDF expression. It also makes it easier to reason *with* >that expression (which I take the second sentence to express). It doesn't >help with social meaning, though, contrary to 4.4: > >"""The meaning of an RDF document includes the social meaning, the formal >meaning, and the social meaning of the formal entailments. The assertion >of an RDF graph G, when G logically entails G', includes the implicit >assertion of G'. The implied assertion of G' should be interpreted using >the same social conventions that are reasonably used to interpret the >assertion of G.""" > >My reaction to this is, "Wow, that's *CRAZY*!!!!" I do not exaggerate. >There's also a bunch of inconsistency in that section, for example: > >4.2: """Such an assertion should be understood to carry the same social >import and responsibilities as an assertion in any other format.""" > >But the normal social import and responsibilities of an assertion in HTML >is *not* the social import and responsibilities of the *formal* >entailments (especially not the FOL entailments of an RDF regimentation >of) that HTML. (For one, I'd rather be committed to the relevant >implications of my statments.) Plus, consider the common social meaning of >Moore's paradox: "I don't believe it's raining". > >""""A combination of social (e.g. legal) and technical machinery >(protocols, file formats, publication frameworks) provide the contexts >that fix the intended meanings of the vocabulary of some piece of RDF, and >which distinguish assertions from other uses (e.g. citations, denials or >illustrations).""" > >Well, in a *normative specification*, I would expect this combination to >be spelt out. It isn't. > >Plus, *intended* meanings, by their nature *ALWAYS* can come unstuck from >the conventions and formal/techincal structures of the meaning carrying >expression. How does speaker's meaning (vs. sentence meaning) come into >play. If I *intended* to refer to John when I say Mary, what have I done >from the RDF point of view? > >There's a large philosophical literature on this. At best, you can hope to >fix the *sentence* meaning. > >"""The social machinery includes the form of publication: publishing some >unqualified statements on one's World Wide Web home page would generally >be taken as an assertion of those statements. """ > >Generally. And if not so taken? Is it *correct* to so take? On what >ground? In which juristiction? How is this normative? > >"""But publishing the same statements with a qualification, such as "here >are some common myths", or as part of a rebuttal, would likely not be >construed as an assertion of the truth of those statements. Similar >considerations apply to the publication of assertions expressed in RDF.""" > >Be construed. By whom? Speaking as someone who's *often* misread (e.g., as >asserting things I explicitly deny), what other people *construe* of what >I assert is very far from determinative of what I assert. > >"""An RDF graph may contain "defining information" that is opaque to >logical reasoners. This information may be used by human interpreters of >RDF information, or programmers writing software to perform specialized >forms of deduction in the Semantic Web.""" > >Again with the scare quotes. Scare quotes are used to indicate that the >enclose words are not being used in their normal sense. But there's no >vague, much less precise, definition of "defining information". And I'm a >logical reasoner, will this information be opaque to me? (Well, if in >German, yes, but *all* human reasoners?) > >***4.3: > >""The social conventions surrounding use of RDF assume that any RDF URI >reference gains its meaning from some defining individual, organization or >context. This applies most notably to RDF predicate URI references.""" > >Actually I presume that people do the presuming. And are these social >conventions true in all societies? Are you reporting, or specifying? > >""""Thus, the choice of terms used in published RDF is significant in >determining its meaning, through reference to definitions asserted by the >defining authorities for those terms.""" > >How does this fit in with my ability to use any URI in my ontology, making >any defining assertions I like about it? > >""""However, even when a URI reference can be dereferenced as an RDF/XML >document, it's use within an asserted RDF graph does not implicitly assert >the contents of the referenced document."""" > >So it's formal meaning isn't fixed IN ANY WAY by the "authority"? And the >social meaning? > >***4.4 > >"""Human publishers of RDF content commit themselves to the >mechanically-inferred social obligations."""" > >I discussed this above. It's also unclear what the mechanically-inferred >*social obligations* are (as opposed to the mechanically inferred *formal >entailments*). > >"""The meaning of an RDF document includes the social meaning, the formal >meaning, and the social meaning of the formal entailments. The assertion >of an RDF graph G, when G logically entails G', includes the implicit >assertion of G'. The implied assertion of G' should be interpreted using >the same social conventions that are reasonably used to interpret the >assertion of G.""" > >Oh, this might fix it. But what if the social meaning of the set of >explicit assertions is different from the set of those of the set of >implicit assertion? heck, if people general work with a relevance logic >(more likely than RDF entailment), then you *explicitly* don't always >endorse/believe P & ~P even if you assert/believe each conjunct separately. > >Finally, if I assert a contradiction (which I can do, *even in rdf*), am I >thereby commmited to the social meaning of EVERYTHING? > >Oops, out of time. > >Cheers, >Bijan Parsia. ------------------- Graham Klyne <GK@NineByNine.org>
Received on Tuesday, 25 February 2003 12:19:45 UTC