W3C home > Mailing lists > Public > semantic-web@w3.org > March 2011

Re: thisiness

From: Henry Story <henry.story@bblfish.net>
Date: Fri, 25 Mar 2011 17:16:48 +0100
Cc: nathan@webr3.org, Gregg Reynolds <dev@mobileink.com>, SW-forum Web <semantic-web@w3.org>, Halpin Harry <hhalpin@w3.org>
Message-Id: <F9AAE9DA-0507-477A-8B4C-AF1DF49D376B@bblfish.net>
To: Pat Hayes <phayes@ihmc.us>

On 25 Mar 2011, at 15:41, Pat Hayes wrote:

> 
> On Mar 25, 2011, at 7:51 AM, Henry Story wrote:
> 
>> 
>> On 25 Mar 2011, at 13:36, Nathan wrote:
>> 
>>> Henry Story wrote:
>>>> On 25 Mar 2011, at 12:38, Gregg Reynolds wrote:
>>>>> (BTW, there are no unlabeled nodes; what seem to be unlabeled nodes are in fact anaphorically or indexically labeled.  An unlabeled circle in a diagram is still a symbol, a kind of indexical label; where nesting is used syntactically to indicate a blank node, the syntactic device of nesting can be taken as a kind of indexical label or symbol.  In other words, the difference is not between labeled and unlabeled, but between persistently and transiently labeled (or: between rigidly and contigently labeled) nodes.)
>>>> yes, that is what I was getting at in my last post. Blank nodes allow non rigid designation. URIs tend to be rigid designators. Somehow in philosophy this distinction seems to have been very helpful. It would be interesting to follow through on this parallel and see if we don't find the same usage and value on the semweb.
>>> 
>>> We have the term "blank node identifiers" which leads many to believe that the identifier names the blank node, but there is no scope for that name, so they are non-rigid designators.
> 
> Um...no, this muddles different topics. They are non-rigid designators, indeed, but not because of scope issues. (Because of their existential semantics, is the reason.) 
> 
>>> The problem is that now often many see certain sets of triples as being associated with a name somehow  (as in sparql, the ?g, and linked data, the <uri> you GET) which gives a scope for the name, and thus the expectation is that blank node identifiers are rigid designators within the scope of that name. (As in, a name which acts as a namespace and defines a universe within which blank nodes can be both quantified and identified/named).
>>> 
>>> Blank nodes are pretty much existentially bound variables with a quantifier whose scope is the entire graph, so when you have a name for two graphs and the name is the same, it follows that they are thought of as bound variables rather than free variables.
>>> 
>>> Even if the above were made so in some way, we'd arguably end up with even more unexpected functionality when merging graphs associated with different names, and when no name is present.
>> 
>> And part of the reason people identify blank nodes with uris is that the use of the one versus the other has not been made clear. We should work out the pragmatic, philosophical and logical spaces where blank nodes give benefits. 
>> 
>> So if I describe someone using a blank node, then it is not possible to speak of the person using a URL in my namespace.
> 
> Well, it is *possible*. You might for example know that the blank node is owl:sameAs a URI. 
> But I take your point:
> 
>> This removes a certain identification responsibility from me. 
> 
> Right. But in fact, surely this responsibility resides with the *owner* of the URI rather than the user of the URI. 

That was my hypothesis. 
If inside http://bblfish.example/doc I wrote

[] a foaf:Person;
   foaf:name "George" .

Then people will not be able to use an identifier I coined to speak about said George. They will have to identify George by description, which is very different from using a URL in my namespace. With blank nodes I can be sure I am speak of something by description. Now I think we should work on simple examples where this prooves to be an advantage, inspring ourselves from the work in logic. Perhaps then we will discover the proper way to use blank nodes.

> 
>> 
>> If I remember correctly - Harry Halpin might have this more freshly in his mind - Gareth Evens in "The Varieties of Reference" has a section on the importance of being able to distinguish objects if one wants to name them. He claimed - if I remember correctly - that it is not possible to name any of two identical balls rolling around in a basin, because if one were to be shown any of the balls one would have no way of assigning the name to one or the other of them. Here it seems one can tell that there are two balls in the basin, but not be able to give them names. If this reasoning is correct then there is a fundamental problem with arbitrary skolemisation of blank nodes. 
> 
> No, this does not follow. Skolemizing replaces an existentially bound variable (a bnode) with a logical "name", but this does not imply naming in Even's sense. A logical name need not be a proper name. So, given the two indistinguishable balls in Evens' basin, I can pull a mathematician's trick and say, "call them A and B". This does not "name" either one of them in the sense of being able to identify the particular referent of the name, but it does enable me to say things about the balls in the basin. For example, I can say that A =/= B, since there are two of them. I havn't thereby said anything more than I could have said without using my pseudo-names, of course: I have really only said (exists x, y (x =/= y)), using skolem constants instead of the quantifier. But that is the point: skolem 'names' aren't names in the proper, identifying, rigid identifier sense. They are just brand-new logical constants, whose very newness guarantees that they shall be unemcumbered by any prior burden of referring to anything in particular. Similarly, if we have some URI scheme for use instead of blank nodes, these URIs will not be resolvable and will not provide any information about their denotation other than that it exists.

( Btw. have you  read Evan's book? I am asking because it's a long time I have not read it, and I have always looked for an opportunity to apply this knowledge somehow. It will make a difference in my reading of what you are writing above to know if you have read "The Varieties of Reference". And if you have read it, then it would be helpful to understand how that ties into the semweb.)

If I understand you correctly the proposal is for blank node URIs? Assuming that idea makes sense, what is the difference in how one should use a blank node URI and a normal URI? It would be interesting to know because the answer to that would also then be an answer to how one should use blank nodes - minus the fact that people could now refer to my blank nodes...


> 
> Pat
> 
> 
> 
>> 
>> Henry
>> 
>>> 
>>> 
>>> Best,
>>> 
>>> Nathan
>> 
>> Social Web Architect
>> http://bblfish.net/
>> 
>> 
> 
> ------------------------------------------------------------
> IHMC                                     (850)434 8903 or (650)494 3973   
> 40 South Alcaniz St.           (850)202 4416   office
> Pensacola                            (850)202 4440   fax
> FL 32502                              (850)291 0667   mobile
> phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
> 
> 
> 
> 
> 

Social Web Architect
http://bblfish.net/
Received on Friday, 25 March 2011 16:17:25 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:48:24 UTC