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

Re: GoodRelations Light

From: Alan Ruttenberg <alanruttenberg@gmail.com>
Date: Wed, 18 May 2011 18:34:51 -0400
Message-ID: <BANLkTikq6YL1YtV0un7wJFMp1rSWxM2SdA@mail.gmail.com>
To: glenn mcdonald <glenn@furia.com>
Cc: Michael F Uschold <uschold@gmail.com>, "semantic-web@w3.org" <semantic-web@w3.org>
On Wed, May 18, 2011 at 2:02 PM, glenn mcdonald <glenn@furia.com> wrote:

> things that we haven't given a name. For example when my wife was
>> pregnant, there was a growing embryo that we didn't name for a while.
>> A bnode might be used to represent that growing to-be child.
>>
>
> This, like many examples people give of things you'd use a blank node for,
> misses the point. The embryo may not have a name yet, but it has a gender
> and a conception date and a size and a nucal transparency, is the subject of
> ultrasound imaging, etc. It isn't a logical abstraction, it's a specific
> entity. It should have a URI.
>

I'm not sure how you come to the "should". Whether or not to name something
is a matter of convenience, a human act done for a purpose. I don't object
to having a URI, and in fact my sentiment is similar to yours that given the
choice I would rather have names for entities worth representing in any way,
but that preference is a far cry from being a "should".

I'm not sure to what you are responding when you say: "It isn't a logical
abstraction". I thought I was clear from the start that I was talking about
a physical thing. What sort of thing is this "logical abstraction" and what
is its status in semantic web languages?


> This might sound tangential, but I contend that you'll have a much easier
> time talking about blank nodes clearly if you stick to using them, even in
> examples, only in cases where they're actually required: to state an
> existential quantification, like "somebody must have seen the crash", not
> just to talk about a specific individual for whom we're just missing some
> information.
>

"Somebody must have seen the crash" has more of a modal/defeasible flavor
than an existential quantification. To make you example clear I would
recommend you say instead "Somebody saw the crash". Once you write that in
that way, I would like to know how that statement is any different than
mine. Certainly if somebody saw the crash, there is a specific individual
(at least) who saw the crash. Other statements might even add additional
constraints leading to the conclusion that there is only a single individual
who saw the crash.

Perhaps you are trying to say that use of a blank node implies some
epistemic status - it reflect some level of knowledge we have. You are
seeming to say: use a blank node if you know very little about the entity.
Unfortunately this interpretation is not at all supported by the
specifications or the logic. I would worry that such a presentation of blank
nodes would mislead the audience.

Once again: a blank node is a representation of some thing that the
representing agent has chosen not to give a name to.

(But then, I also contend that this whole concept should be moved out of RDF
> into OWL.)
>

While as a technical issue this might make sense, as an operational issue it
is a very bad idea. I don't see that any backwards incompatible change to
RDF would justify the cost in fixing tooling built to the current
specifications, and making such a change would be damaging to the reputation
of the W3Cs standards-making efforts.

-Alan

-Alan
Received on Wednesday, 18 May 2011 22:35:39 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:42:27 UTC