Re: Landscape Notes

From: Sandro Hawke <sandro@w3.org>
Subject: Landscape Notes
Date: Tue, 23 Sep 2003 14:05:52 -0400

[...]

> I have had more luck coming up with a list of what might be called
> "issues", but might also be called "landmarks", because they are more
> like common threads to the debate than a typical WG issues list entry.
> 
> But anyway, here's what I've got so far....  1-9 are IMO in scope and
> 10-13 are IMO out of scope.
> 
>      -- sandro
> 
> ================================================================
> 
> These are not simple issues and they are not likely to be closed to
> everyone's satisfaction in the near future, if ever. They are
> instead landmarks helping us navigate the space of decisions that
> could be made in this area.  Each "issue" seems to be the focus of
> a rough cluster of possible decisions and a common thread of
> discussion.  Knowing the landscape may help us avoid getting lost,
> and lets us study the problem space in more manageable chunks. 
> 
> The names of the issues are necessarily imperfect.  Think of them
> as mere mnemonics.
>    
> Feedback welcome, of course.
> 
> 1.  What-Is-Identifies
> 
>     What is the relationship between a URI and the thing (or things)
>     it "identifies"?
> 
>     Pat/TimBL - does a URI denote just one thing?   Agreed that URIs
>     "work the same" everywhere, or something.
> 
>     Issuette: Slipperiness of Intended Meaning
>     http://www.w3.org/2001/sw/meetings/tech-200303/social-meaning#fish
> 
>     In what contexts are multiple occurances of URI defined to
>     mean the same thing"
> 
>     Does an URI mean the same thing in every HTML HREF?  Every
>     rdf:about/rdf:resource? The same in both?

I would hope that we can come up with a formal answer to this and leave
the slipperinesses alone.

> 2.  Use-Implies-Consent
> 
>     Does use of a URI as a name in RDF consititue some kind of a
>     commitment or consent, either to meaning expressed on the web, or
>     intended by a Name-Authority (below)?
> 
>     Recent stuff by PFPS.   Consent/Committment/Something.
> 
>     This is the most-attacked bit of "Tim's Proposal", whatever that
>     is. 

I think that this is the central issue that needs to be addressed.  In
particular, what can replace the proposed totalitarian solution to this
issue?

> 3.  Name-Authority
> 
>     Who (if anyone) gets to say what a URI means?
> 
>     Recent stuff by Bijan.  
>     
>     Does someone get special status when to say something about
>     something when a particular name is used for it?
> 
>     issuette: Authoritative Definition of URIs
>     http://www.w3.org/2001/sw/meetings/tech-200303/social-meaning#authority

This needs to be addressed.  Fortunately, RDF has already mandated a
solution.  :-)  (The freedom to say ``anything about anything'' is also the
freedom to not listen to what the defining authority has said!)

> 4.  Communicating-Meaning
> 
>     How does one express or convey what a URI means?
> 
>     issuette: Defining Information
>     http://www.w3.org/2001/sw/meetings/tech-200303/social-meaning#defn
> 
>     How can a user learn about something, given a URI for it

Hmm.  This appears to be a bit unfocused.  I would say that there are at
least three related notions here: how and where information is stored (at
URIs, in documents written in a SW language), how to indicate that a
particular document is to be used (explicit import directives), and what
best practices might say is reasonable in the absence of other guidance
(strip off the fragment identifier and look there for a document in a
SW language you understand).

> 5.  What-Is-Representation
>     
>     How does a thing relate to its HTTP 'representations'?
> 
>     What exactly is the "representation" relationship (the
>     relationship between the thing a URI "identifies" and the MIME
>     Entity/bytestring retrieved as its "representation") ?

I would morph this into

	Which document/document section gets retrieved when a URI reference
	is dereferenced?

Other aspects are, I hope, beyond our scope.

> 6.  Centrality-Of-Predicates
> 
>     Do URIs in the predicate position of an RDF triple have some
>     special role in the meaning of the triple?

Needs addressing, if only to categorically deny.  :-)

> 7.  Names-As-Web-Addresses
> 
>     Can a URI work as a name in RDF and a web address, giving good
>     content in a normal browser?  If so, how?

I would be happy not addressing this one.

> 8.  Layering
> 
>     OWL has multiple "levels" (Lite, DL, and Full) which have
>     slightly different semantics.  An RDF graph means slightly
>     different things depending on which level is intended; is
>     this okay?   

Consider also RDF vs RDFS.   I believe that this one needs to be addresssed.

>     More generally, when (if ever) can the meaning of a term be
>     safely extended by another layer?

This is a separate issue that I don't think we need to address.

> 9.  Change-Over-Time
> 
>     What happens if the Name-Authority changes its mind or what it
>     says? 

I hope that we don't have to address this at this time.  

> 10. What-Is-Asserted
> 
>     (Out of scope for tag-39; responsibility issue)
> 
>     How can you tell which RDF you see is asserted to be true?
> 
>     http://www.w3.org/2001/sw/meetings/tech-200303/social-meaning#assert

Hmm.  Perhaps not out of scope, but I hope that we only need say that for
our purposes if it is there, then it is (formally) asserted.

> 11. Commitment-To-Implications
> 
>     (Out of scope for tag-39; responsibility issue)
> 
>     If someone commits to some RDF, are they also committing to its
>     entailments in some logics?   Which logic?
> 
>     http://www.w3.org/2001/sw/meetings/tech-200303/social-meaning#assert

Committment is out of the scope here, I think.

> 12. Who-Makes-Committment
> 
>     (Out of scope for tag-39; responsibility issue)
> 
>     If RDF is asserted, perhaps by just being on the web, how can
>     you tell who is asserting it?  Who is ultimately responsible?
> 
>     http://www.w3.org/2001/sw/meetings/tech-200303/social-meaning#third
> 
>     Signatures may be help here.

Also out of scope.

> 13. Affecting-Law
> 
>     (Out of scope for tag-39; responsibility issue)
> 
>     Should laws be changed to support communication via the semantic
>     web?  Should people be given guidence as to how we think laws
>     should be interpreted in considering semantic web issues?
> 
>     http://www.w3.org/2001/sw/meetings/tech-200303/social-meaning#law

Definitely out of scope.

peter

Received on Wednesday, 24 September 2003 09:15:13 UTC