W3C home > Mailing lists > Public > www-archive@w3.org > November 2012

Re: Working without being ambushed by Ambiguity

From: Graham Klyne <GK@ninebynine.org>
Date: Thu, 01 Nov 2012 22:01:07 +0000
Message-ID: <5092F123.5060403@ninebynine.org>
To: Larry Masinter <masinter@adobe.com>
CC: Pat Hayes <phayes@ihmc.us>, Tim Berners-Lee <timbl@w3.org>, David Booth <david@dbooth.org>, "www-archive@w3.org" <www-archive@w3.org>
On 01/11/2012 08:44, Larry Masinter wrote:> (cc to www-archive, not www-tag)
 > I want to take this topic off www-tag, ...

Good.  Maybe I can be a bit less guarded in what I say?

TL;DR:  I think many of our difficulties here would be eased if the RDF working 
group would bless a semantic framework for using RDF Datasets.

...

For me, there are several intertwined issues here, in no particular order:
- context
- ambiguity
- vagueness
- sound inference
- modalities (? - I mean conflicting or differing interpretations in a common 
discourse)

What we *have* in the present model theoretic approach is sound inference.  In 
particular, with RDF, the idea that the RDF merge of two (or more) graphs is 
true under exactly the interpretations that make the original graphs true.  I 
think this is a key necessity (but not sufficiency) for combining and remixing 
data on the web through automated processing, and of itself represents an 
important step forwards from what we had before.  I'm reluctant to let that go.

Along with this, I think vagueness is somewhat covered by a Quine-like appeal to 
consideration of statements that people broadly accept as true, if one doesn't 
get too hung up on exactly *what* is denoted by individual terms, just accepting 
that they have denotations that satisfy certain properties.

I think that ambiguity of the kind that permits Herbrand style models is 
something that we should just ignore - it seems to me that trying to exclude 
this kind of ambiguity in the formal structures leads to the kind of tar-pit 
we've been wading in.

I *think*, BICBW, the last two points somewhat reflect what Tim was trying to 
say in his original "without being ambushed by Ambiguity" - so to that extent we 
may agree.

But what we don't have is a satisfactory, easy to follow story that covers 
context and modality (if "modality" is the right word to use here).  Which would 
(should) extend to topics like "slander".

Here, I fear we're being let down by the RDF working group.  They have agreed a 
structure, RDF Datasets, that is capable of encoding such ideas, but seem unable 
to come to a consensus on how to provide semantic underpinning for using this 
structure.  IMO, *any* semantic underpinning would be better than none - without 
it, we're back in the mess we had figuring reification last time round.  (What I 
was hoping for is *not* a definitive "this is what datasets mean", but a 
framework within which one could construct semantics for datasets without fear 
that the ground would later shift.)  There have been several proposals, and at 
least two that I'm aware of in the life of the current RDF group - including 
Pat's RDF as context logic - any (or most) of which could serve.

(Personally, I liked the proposal that was made, and apparently rejected, a 
month or so ago 
(http://www.w3.org/2011/rdf-wg/wiki/TF-Graphs/Minimal-dataset-semantics).  I 
have the impression, maybe wrong, that Pat's context logic approach was a bit 
more constrained, but still flexible enough to support a useful range of 
modalities.)

Given this much, we would have some basis for actually talking about (or 
representing) some of the tricky issues that are so hard to discuss in the 
current "one interpretation to rule them all" view of RDF (and URIs).  We could 
propose structures that capture belief, provenance (which I come to see can 
itself be highly contextual), disagreement, debate, conditionality, and so much 
more.  Maybe then we also have a framework for encoding the theory of speech 
acts, etc?

If we have a way to represent and talk about contextualization, then I think the 
whole issue of a URI having different interpretations in different contexts (or 
applications) is something we can accommodate.  That is, it allows us to set out 
without a presumption of global meaning, yet still exploit the commonalities we 
can observe.   Within RDF as we currently have it, we're forced to go "out of 
band", and that makes it hard to really understand each other's difficulties.

...

As for "attrition", I don't think we're dealing with a belligerent enemy here.

But I do feel like I'm on the rough edge of the grindstone here.  For the most 
part, I can ignore this stuff in my daily work with RDF:  99% of the time it 
seems it just doesn't matter.  But I fear if we don't build on sound foundations 
then sooner or later things will start to crumble.  I care if that's the case, 
but a lot less than I care about a lot of other things, so my forays into this 
arena will be of limited energy.  Maybe that's for the best.

#g
--


On 01/11/2012 08:44, Larry Masinter wrote:
> (cc to www-archive, not www-tag)
>
> Wikipedia: Attrition warfare is a military strategy in which a belligerent side attempts to win a war by wearing down its enemy to the point of collapse through continuous losses in personnel and matériel."
>
> If you want an alternative to model theory, start with speech acts, Searle, Winograd & Flores. You don't have to start with global meaning and add on contexts.
>
> I want to take this topic off www-tag, though, since it causes people to leave the TAG or not to consider joining the TAG.
>
> Larry
>
>
>
>
>> FWIW, I agree with you about contexts. I have a fairly detailed proposal to make
>> RDF into a context logic, which could be done without affecting existing
>> deployment and would resolve, or at any rate ease, a number of contentious
>> issues, including http-range-14 and the best way to interpret SPARQL
>> datastores. But it hasnt got any traction partly because people find model
>> theoretic thinking too difficult, apparently. You can see it at
>> http://www.slideshare.net/PatHayes/rdf-with-contexts (meat starts at slide 10)
>
> I think that's a start, but it doesn't go far enough, because it doesn't talk about trust, belief, provenance (except perhaps in a very simplistic way).
>   And most people stop at what you call the globalist model and get stuck there.
>
>>>> We need to get the naming and the sentential structure clear before we can
>> begin to make sense of ideas like provenance, belief or trust.
>>>
>>> I disagree that doing so is a "need". I understand that is the approach you're
>> taking, and it isn't working.
>>
>> Its working pretty well. It handles all the inference-making on the Web, for
>> example.
>
> Cf "War of Attrition" above. It isn't working for the TAG. I don't know who it is working for.
>
> A challenge: try to explain any of the use cases of http://www.w3.org/2001/tag/doc/publishingAndLinkingOnTheWeb.html
>
> using contexts. Explain 'slander'.
>
>>> I don't think the model which starts with reference and adds provenance,
>> belief, and trust, is adequate for the kind of analysis needed.
>>> I'm looking for a model that starts with reference being a personal individual
>> act, and communication is based on correlation through turst.
>>
>> If reference is a personal individual act, how do I communicate my references to
>> you?
>
> Will my point be more clear if you change
>       s/personal individual act/local context/g
> in your model?
>
>>   It is exactly the communication that we need to analyze. And if
>> communication is correlation, then correlation *of what*? And how is such
>> correlation achieved? Presumably, by sending messages of some kind. If you
>> pursue this line of thought, I bet you will quite quickly re-invent Tarski's theory
>> of truth. Communication is achieving alignment of beliefs by transmitting
>> sentences expressing propositions. That is exactly what model theory is about.
>
> If it were "exactly" what model theory was about, you wouldn't have to do all this extra work.
>
>>>>> For example, the notion of "meaning" is tied completely to persistence.... a
>>>> use of a URI is "cool" to the extent that its effect on recipients (its
>> "meaning")
>>>> doesn't change (over time or the range of recipients).
>>>>
>>>> Wrong on several counts.
>>>
>>> I put "meaning" in quotes because I wasn't using the term in a way I thought
>> you would agree with. Rather than argue about it, just try it out as a definition
>> instead, change "is tied" to "should be tied".
>>
>> Well, first, why don't you tell me what *you* mean by "meaning" ? Start with a
>> simple example like "Snow is white." :-)
>
> I'm not interested in trying to define "meaning" or in theories or an architecture of the web that requires one. So I think it's a rathole, that we don't need to define or use "meaning" "denote" or "identify", or treat "http" as a "function".
>
>>>> Everest is a mountain, what I mean is something about a place in Nepal, and
>>>> mountains. I have no idea what a recipient is going to use this for or what effect
>>>> it might have on them, and that has nothing to do with what I mean.
>>>
>>> Humpty Dumpty used words without any idea of what a recipient might use it
>>> for or what effect it might have on them.
>>
>> Humpty Dumpty said that words mean what he wanted them to mean, because
>> he was the master. That sounds to me more like your idea that reference is a
>> purely private matter.
>
> OK, you can sling eggs.
>
>>> I think you're doing yourself a disservice by claiming you have _no_ idea. You
>> don't have a complete and accurate idea. When you want to say something, you
>> say enough, with enough hyperlinks and other reference material, to help the
>> poor reader.
>>
>> But suppose that I have no idea who the reader is, or what use they are going to
>> make of my words? All I can presume is, they understand English (or RDF, etc.).
>> And surely that is the normal case on the Web. The hyperlinks and reference
>> materials help the reader better understand what propositions I am trying to
>> say, sure. But as to *why* the reader might be interested in this stuff, or *how*
>> it might influence their behavior, I neither know nor care. We communicate at
>> the level of transmitting propositions, but it doesn't get more intimate than that.
>
> You're saying this as if you believe it is "fact" rather than just one of several models.
>
>>>> (And this is
>>>> just as well, since on the Web I cannot *possibly* know what effect it might
>>>> have on them.)
>>>
>>> Of course, you can't possibly know EVERTHING about what will happen. But
>>> surely it is the writer's obligation to consider the reader.
>>
>> But not to know so much about the reader that one is setting out to sculpt their
>> behavior, or still less that this sculpting is the meaning of ones words.
>
> If you want a theory, start with Austin/ Searle / Speech acts.
>
>>>
>>>> Second, meaning does not depend upon persistence.
>>>
>>> Insofar as communication travels the speed of light or lower, there is a time
>> delay between the writer writing and the reader reading. Sometimes it's
>> minutes, sometimes days, weeks, months, years, decades, centuries.   Over
>> many of those time scales, the web changes. Insofar as you intend future
>> readers to understand what you write, you use terms that will retain some
>> quality. Perhaps you don't mind if a blogger changes the style of their blog, as
>> long as the content is still there. The things you don't mind changing are
>> irrelevant to meaning, the things you mind if they change are highly correlated
>> with meaning of the link.
>>
>> All that means is, you try to keep your meaning expressed in a way that will last.
>> For example, its not a good idea to use tenses without putting a date on your
>> writing so that readers can figure out when you are talking about. But all of this
>> rests on the fact that your writing is expressing propositions. If you were writing
>> music, for example, none of this would apply. You would just write the notes on
>> the stave, and that would be that. No dates, no trust, no content (other than
>> the music itself), and no real communication; all because, no propositions.
>
> You have too narrow a view of communication, IMO.   Writing notes on a stave
> comes with some expectation of behavior of performance, and each note has
> a meaning.
>
>>>> If my
>>>> uncool URI changes its reference every week, then each week's meaning is
>>>> perfectly clear (we can suppose) even though persistence is lacking.
>>>
>>> See time delay above.
>>
>> I wasnt talking about that issue. My point was, that even when there is no
>> persistence, there can still be perfectly clear meaning. So the latter can't depend
>> upon the former.
>
> Nothing is "perfectly clear", "clear" is a matter of degree and never reaches
> "perfect". And whatever clarity you have generally decays over time, but
> It's hard to imagine a use case where the duration of clarity is infinitesimal.
> So no, I think  there are no cases where there is "no persistence" and there is
> "perfectly clear meaning".  If  you think you have one, please explain it.
>
>
>
>>>> Weekly
>>>> magazines might be good examples. Persistence might in fact be better defined
>>>> in terms of reference and meaning rather than the other way around: for
>>>> example, RDF and SPARQL have spend some time defining notions of
>>>> equivalence which allow referents to change 'harmlessly' precisely because the
>>>> *meaning* doesnt change.
>>>
>>> Exactly, I just think you are looking in the wrong end of the telescope. Turn it
>> around. Start with persistence, and tell me what you see about meaning.
>>
>> Nothing much, for the reason above.
>
> Try harder.
> f
>>>>> As long as we talk about statements (triples) or terms (uris) having
>>>> disembodied meaning or fail to take into account trust, error and malice,  we're
>>>> not making progress....  and wasting time.
>>>>
>>>> Until we do get "disembodied" meaning clear, we will never get out of the can
>>>> of worms far enough to even get started on these harder problems. I agree
>>>> they aren't afterthoughts, but they are thoughts that must come after.
>>>
>>> If you start with what you think is the beginning, then yes, other thoughts
>> must come after. I'm suggesting a different starting point, since the one you've
>> been using isn't working.
>>
>> Well, as I say, I think it is working. Progress is slow, but these are complicated
>> issues which havnt yet caught the attention of most of the relevant specialists.
>> Model theoretic semantics has a long history and has been applied to a very
>> wide range of kinds of language and communication. At the very least, it makes
>> sense to adapt it to the Web rather than throw it away and try to re-invent the
>> wheel.
>
> There are other wheels. Try one out, this one isn't helping.
>
>>
>> (**) OK, there is a notion of trust where what I trust is that you will *do* some
>> predictable thing in response to input from me. In this sense I trust my cat to
>> look at me when I call his name, and my OS to open a menu when I click on the
>> desktop background. And this doesnt reat upon communication of propositions.
>> But I don't think that the sense of trust involved in Web thinking is this simplistic.
>
>
>
>
>
Received on Thursday, 1 November 2012 22:58:03 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 7 November 2012 14:18:59 GMT