Re: [sw-meaning] Agenda & Logistics for Tomorrow's Telecon

At 15:00 30/10/03 -0500, Sandro Hawke wrote:
>     Pat Hayes writes:
>     http://lists.w3.org/Archives/Public/public-sw-meaning/2003Oct/0092
>
>         Words convey meaning to us humans, and we all use them to
>         convey meanings to others. But this works, often enough and
>         well enough to be useful, not because the meanings that the
>         words have for speaker, and those that they have for the
>         hearer, are *identical*, still less that there is a single
>         unique such meaning; but rather because the people involved
>         have enough of an overlap in their conceptions that the hearer
>         is able, using the surrounding words and the nonlinguistic
>         context of the conversation, to extract enough of the
>         speaker's intended meaning for the communicative purpose which
>         happens to be relevant at the time.
>
>     and
>     http://lists.w3.org/Archives/Public/public-sw-meaning/2003Oct/0093
>
>         > How is a URI different from a constant?        (James Lynn)
>
>         For these discussions, not a lot. RFC 2396 insists that any URI
>         "identifies" a unique resource, but it provides no way for a 
> resource
>         to be baptized by a URI, and a name that isn't attached to any
>         referent is hardly distinguishable from a constant symbol that isn't
>         a name.

While I agree with these messages, I'm not sure how they help with respect 
to dealing with the full spectrum of meanings that might be attached to SW 
expressions.  Except, maybe, from dissuading us from talking about the 
meaning of URIs?

Is there some analogue for "nonlinguistic context of the conversation" that 
might be applicable to SW agents?  Possible candidates might be:
- assumptions built-in by the programmer
- application conformance to non-formal statements in imported documents
- acceptance as truth statements some given oracle(s)
- conforming behaviour over given test suites
   :
   ?

#g


------------
Graham Klyne
For email:
http://www.ninebynine.org/#Contact

Received on Thursday, 6 November 2003 09:12:40 UTC