W3C home > Mailing lists > Public > public-rdf-comments@w3.org > October 2013

Re: RDF Semantics - Intuitive summary needs to be scoped to interpretations

From: David Booth <david@dbooth.org>
Date: Thu, 03 Oct 2013 23:55:27 -0400
Message-ID: <524E3C2F.9050303@dbooth.org>
To: Pat Hayes <phayes@ihmc.us>, public-rdf-comments <public-rdf-comments@w3.org>
[Resending, copying the public-rdf-comments list]

Hi Pat,

On 10/03/2013 10:32 PM, Pat Hayes wrote:
> David, greetings.
>
> This is a response to your email, copied below, which raised
> issue-149 [1].
>
> The section in question is quite explicit that it is intended to
> provide an intuitive, non-technical description of what the formal
> semantics describes exactly. The intuition in question is that graphs
> have a truth-value which is completely determined by what their
> component IRIs denote.

Right, but "what their component IRIs denote" is determined *solely* by
the interpretation that is applied.  You cannot leave out or gloss over
the importance of interpretations.  They are absolutely central to the
RDF Semantics.

> This is also exactly the intuition underlying
> the formal model theory, reflected formally by the fact that an
> interpretation is a function on all IRIs, and that this function
> completely determines the meanings of all RDF structures larger than
> a single IRI.

Yes, exactly.  This notion of an interpretation is absolutely central to
understanding the RDF Semantics.  An intuitive introduction must not
omit or gloss over it.

> The notion of "interpretation" is a mathematical device
> for expressing the fact that the denotations of IRIs intended by the
> writer of some RDF may not be known by the reader, and is used to
> account for the fact that valid inference rules are immune to such
> communicative disconnects (since validity means that it holds
> regardless of which interpretation is applied to the IRIs.)

[Hmm, not relevant to this issue, but it's interesting to see that you
have noted that the denotations of IRIs intended by the writer may
differ across writers.]

>
> The avoidance of the formal "interpretation" language in this
> informal, descriptive, account is deliberate,

That's a big mistake, because the notion of interpretation is so central
to the formal semantics.  Readers cannot possibly understand the
semantics without understanding the notion of an interpretation.

> as is the use of the
> verb "interpret" to suggest how the formal mathematical account given
> in the surrounding normative text should be understood to relate to
> this intuitive, informative, summary. The interpretations mentioned
> in point 4 are indeed the results of the verb "interpret" in point 2,
> as you surmise. Readers who see the connection should not be confused
> by this confluence of naming, which is deliberate. One could describe
> an interpretation as a "way of interpreting" what IRIs mean.

That's a reasonable way of describing an interpretation.  But point 2 is
talking about something different: "2. there is some way to interpret
all the blank nodes in the graph as referring to things".  Mappings from
blank nodes to things are *not* called interpretations in the formal
semantics, they are merely called mappings.  So I do not see how point 4
can possibly be talking about the same notion of interpretation as point
2 and *also* be talking about the formal semantics' notion of
interpretation.  So I am still confused about what notion of
"interpretations" is intended in point 4, which says: "4. and under
these interpretations, each triple S P O in the graph asserts that the
thing referred to as S, and the thing referred to as O, do in fact stand
in the relationship referred to by P."

How about the following wording instead:
[[
An RDF graph is true under a given interpretation exactly when:

1. the IRIs and literals in subject or object position in the graph all
refer to things,

2. there is some way to map all the blank nodes in the graph to things,

3. the IRIs in property position refer to binary relationships, and

4. each triple S P O in the graph asserts that the thing referred to as
S, and the thing referred to as O, do in fact stand in the relationship
referred to by P.
]]

>
> Your suggested re-wording would re-introduce the mathematical
> terminology into the intuitive summary, defeating its purpose.

But this document is about the formal semantics, and the notion of an
interpretation is absolutely central to understanding the formal semantics.

> More
> seriously, it is not in fact correct to say that a graph is true when
> *there exists* an interpretation, etc.. That wording defines the
> notion of satisfiability rather than truth.

Quite true.

>
> For these reasons, we have decided not to change the wording of
> section 5.2 in the way you suggest.
>
> Please reply to public-rdf-comments@w3.org indicating whether this
> reply is an adequate response to your comment.

No, this is not acceptable.  I'm sure the first wording that I suggested
could be improved, but interpretations are absolutely central to the
formal semantics and need to be included in the intuitive introduction.
  Please consider the revised wording suggested above, or some other
wording that properly scopes this introduction.

Thanks,
David

>
> Pat Hayes (for the RDF WG)
>
> [1] https://www.w3.org/2011/rdf-wg/track/issues/149
>
>
> On Oct 2, 2013, at 12:15 AM, David Booth wrote:
>
>> https://dvcs.w3.org/hg/rdf/raw-file/default/rdf-mt/index.html
>>
>> Section 5.2 Intuitive summary needs to be scoped to a particular
>> interpretation or set of interpretations.  At present the
>> interpretations are implicit, and this is misleading because it
>> suggests that the notion of a graph being true is somehow
>> independent of an interpretation, whereas in fact the truth of a
>> graph critically depends on the interpretations that are chosen.
>>
>> I suggest rewording the first sentence of this section from: "An
>> RDF graph is true exactly when: . . . " to: "An RDF graph is true
>> exactly when there exists an interpretation such that: . . . "
>>
>> Also, the verb "interpret" is being used in this clause: "2. there
>> is some way to interpret all the blank nodes in the graph as
>> referring to things,", but that causes confusion with the notion of
>> an interpretation (which is a function).  It would be better to use
>> a different verb at this point.
>>
>> Also point 4 mentions "these interpretations", but it isn't clear
>> what interpretations are meant.  Perhaps it means the results of
>> the verb "interpret" in item 2?  In which case, a different word
>> should be used here also.
>>
>> Thanks, David
>>
>>
>>
>
> ------------------------------------------------------------ IHMC
> (850)434 8903 home 40 South Alcaniz St.            (850)202 4416
> office Pensacola                            (850)202 4440   fax FL
> 32502                              (850)291 0667   mobile
> (preferred) phayes@ihmc.us       http://www.ihmc.us/users/phayes
>
>
>
>
>
>
>
>
>
Received on Friday, 4 October 2013 03:55:54 UTC

This archive was generated by hypermail 2.3.1 : Friday, 4 October 2013 03:55:55 UTC