Re: Proposed minor amendments to RDF 1.1 Semantics §5

On 29 Jul 2017 6:51 am, "Pat Hayes" <phayes@ihmc.us> wrote:

Just to be on the record as editor, I agree with Antoine’s responses here.
Both of Sam's suggested edits are reasonable suggestions, but neither of
them are necessary. No other readers or critics of the documents have found
these passages to be ambiguous or unclear. As the process of changing a
Recommendation is laborious and expensive, these suggestions should be
tabled and brought to the attention of any future WG tasked with revising
the RDF standard.


So, tabled until it can be tabled, then? SGTM

http://www.etymonline.com/index.php?search=table&searchmode=none

Dan



Pat Hayes

> On Jul 27, 2017, at 2:25 PM, Antoine Zimmermann <
antoine.zimmermann@emse.fr> wrote:
>
> Dear Sam,
>
>
> On 27/07/2017 19:00, Sam Kuper wrote:
>> RDF 1.1 Semantics §5 says[1]:
>>
>> "IEXT(x), called the extension of x, is a set of pairs which identify
>> the arguments for which the property is true, that is, a binary
>> relational extension."
>>
>> I propose two small amendments to this sentence.
>>
>>
>> # Proposed amendment 1
>>
>> s/pairs/ordered pairs/
>
> In most mathematical writing, it is common to call a pair something that
has a first element and a second element. So the pair (x,y) (or "< x, y >"
as written in the recommendation) is not the same as the pair (y,x) because
the first elements are not the same. In other term, a pair is most often
denoting an ordered pair. The notion of "ordered pair" is often used when
it is important to distinguish it from an "unordered pair" which is in fact
a set of two elements {a,b}. Similarly, an RDF triple (s,p,o) is an
"ordered triple".
>
> In any case, even if your notion of a pair differs from this one, the
formal definition of IEXT is quite explicit about what it means because
IEXT is a mapping "from IP into the powerset of IR x IR" and the notions of
powerset and Cartesian product have very precise and well accepted meanings.
>
>>
>> Why? Suppose this triple is true, and is the entire graph:
>>
>> ex:a ex:x ex:b
>>
>> Also suppose that triple, which happens not to be part of the graph, is
false:
>>
>> ex:b ex:x ex:a
>>
>>> From the first one, we could say, IEXT(ex:x) = {(ex:a,ex:b)}.
>>
>>> From this in turn, if (ex:a,ex:b) is not an ordered pair, then we
>> could conclude that the following triple is true:
>>
>> ex:b ex:x ex:a
>>
>> However, it is not true (see above). So, we must consider (a,b) to be
>> an ordered pair.
>>
>> If I am mistaken about this, I would be grateful for an explanation as
>> to the nature of my error.
>>
>>
>> # Proposed amendment 2
>>
>> s/property/property x/
>
> The sentence in which this appears in the section is not a definition,
it's simply explaining what IEXT is representing, in informal terms. I
don't think we need to be overly precise in sentences that are merely
explanatory. Especially since the document is already published as a
recommendation, updates and errata should be made only for actual errors.
>
> Your reading is correct and, at that point of the document, there is no
other properties than x being mentioned, so there is no possible
ambiguities, IMO.
>
>
> Regards,
> --AZ
>
>> Why? Because the sentence should be explicit about which property it
>> is referring to. My reading of the sentence is that x is a property,
>> and x is also the property to which the sentence is referring, but I
>> am not certain that my reading is correct. If it is, then the sentence
>> should be amended as proposed in order to make it clear that that is
>> indeed the correct reading.
>>
>> Equally, if my reading is wrong, then the sentence should be amended
>> differently, in order to clarify the intended reading. (Do please let
>> me know if my reading is wrong!)
>>
>>
>> # Note and valediction
>>
>> If you reply to this email, please CC me, as I am not subscribed to
>> this mailing list.
>>
>> Many thanks!
>>
>> [1] https://www.w3.org/TR/2014/REC-rdf11-mt-20140225/#simple-
interpretations
>>
>
>
>