Re: information resource note #573

On Tue, May 10, 2011 at 9:09 AM, David Booth <david@dbooth.org> wrote:
> On Thu, 2011-04-28 at 18:06 -0400, Jonathan Rees wrote:
>> After struggling for a few days, and rewriting several times, I now
>> have a note on 'information resources' ready for your perusal.
>>
>> http://www.w3.org/2001/tag/awwsw/ir/latest/
>
> Comments:
>
> 1. I think this is a significant step forward, because it is focused on
> the usage goal: to be able to write metadata assertions.  It really
> starts to tie together several of the things we've been discussing for
> months.  I am not sure it will yet be understandable by a general
> audience though.  I think the presentation may need to be improved a
> bit: tightening up terminology, making it more consistent, and
> explaining relationships with existing terms.

This comment is not helpful.

What changes would be needed to make it more generally understandable?

Where is the terminology not tight?

Where is it inconsistent?

> 2. I'm not sure that defining an IR as "something that can be the
> subject of metadata assertions" gives much more clarity than defining it
> as something whose essential characteristics can be conveyed in a
> message.  However, since writing metadata assertions is a key goal, I
> think it does make sense to come at it from that angle.

That's not a definition and I wouldn't have expected anyone to take it
to be, since it only occurs in the abstract. I thought the abstract
could take liberties like this. I'm afraid that if the abstract has to
be rigorous then it will be either long and incomprehensible, or vague
and uninformative.

I'll see what I can do.

> 3. I think it may still be necessary to connect this metadata-oriented
> definition with the Fielding-esque view (which I share) that an IR is
> something that can produce awww:Representations.  In this view it is a
> role in an architectural model.  Hmm, maybe this happens later, in
> section 3.  Maybe a forward reference is needed earlier?

I don't see how this is relevant. The memo is about metadata, not web
architecture. It's supposed to be an asset that it *doesn't* appeal to
any of the webarch fetish - which is what has gotten us into this mess
in the first place.

> 4. I think it is critical that the definition of IR not say that it is
> disjoint with anything.

I don't think it does. Can you be specific as to where you think it does?

> Rather, it should be treated like a marker
> class.
> http://en.wikipedia.org/wiki/Marker_interface_pattern
> However, particular applications may add assertions making IR disjoint
> with other classes.  This document doesn't get into this question
> explicitly at all, but it does come up implicitly, because the AWWW
> definition clearly does suggest some disjointness.  I think it would be
> best to address this head on.

I don't see how this is relevant. It's an explicit goal to avoid
talking about the ontology of information resources - it's an
axiomatic treatment, and the only ontology assumed is "information",
and there is no assumption that a pig is not information.

Anything that's said about IRs or anything else in this document is
supposed to be forced by the goal of explaining Web metadata. If you
find I'm saying something that is not needed in order to explain Web
metadata please point it out.

Perhaps I am taking liberties in using "information resource" the way
I do - but I'm getting sort of tired of this - every time I use my own
word in order to avoid unintended connotations, I get flak because I'm
not using a common word, and then when I switch back to the common
word, I get flak because the things I'm applying the word to seem to
be different from what others expect. At some point I have to stop
trying to please others, because you can't please any individual, much
less more than one.

If there's a serious problem with terminology acceptance it will come
out in later review.

> 5. This statement: "Metadata might be stated of any kind of information
> entity" is sounding closer to the AWWW definition of IR.

In what way is it close? It is actually a tautology - unlike the AWWW
definition, there is no way to argue with it, since it's vacuous.
There is no mention of essential characteristics or messages.

>  It also hints
> at the connection between the metadata view and the Fielding-esque view,
> since a awww:Representation is information.

This has got to be coincidence. I had no intention of so hinting and I
don't want to direct readers to Fielding, since what he says is
completely orthogonal to the subject of the memo. The intersection
point is that if you choose to use a URI to refer to the information
resource at that URI, you had better not also use URIs in the way Roy
promoted before he changed his mind and signed on to httpRange-14.
(Actually I think he continues to be a bit passive-aggressive about
httpRange-14.) I see no reason to make such negative statements here.

> 6. In section 2 paragraph that begins "The same metadata may apply to
> multiple information entities", I think it would be good to mention
> variants, as described in RFC 2616.

Sections 1 and 2 start from first principles. To bring in any mention
of the Web before the time comes would be destructive to the structure
of the memo. Anyhow I don't see what this would explain.

Maybe in the discussion of HTTP more could be said about connections
to what the RFC says, but I really prefer to exclude any information
that's not essential to getting the point across.

If this was a scholarly paper there might be a "discussion" section
that goes through all the related literature. But that was not my
intent and I would not look forward to researching such a monster.
(See http://www.w3.org/wiki/HttpRange14Webography .)

> 7. The diagram in section 3 uses the term "generalizes" where
> "hasRepresentation" might also appear.  Somehow the terms need to be
> related.

How do you define "has representation"? The word is so ambiguous, and
the relationship is so poorly specified, that bringing it into the
memo would be seriously confusing.
http://odontomachus.wordpress.com/2011/03/07/are-you-confused-yet-about-the-word-representation/

Avoiding "has representation" was an explicit goal for the memo.

> 8. Minor suggested edits:
>
> s/rhetorical practice/practice/
>
> s/The note/This note/
>
> Change: "First, the idea of generic entities that have metadata is
> introduced without any particular reference to the Web."
> to: "First, without any particular reference to the Web, it introduces
> the idea of generic entities that have metadata."
>
> s/where metadata that does not name/where metadata does not name/
>
> s/Eliabeth/Elizabeth/
>
> Regarding "if G generalizes S": it would be good to say what S and G
> are.

I don't think it matters, formally, and it's a formal statement. Isn't
it clear that they're universally quantified? If not I'll add the
universal quantifier "for any G and S".

Jonathan

> Regarding 'G is "on the Web" at U': again, it would be good to say what
> G is.
>
>
>
> --
> David Booth, Ph.D.
> http://dbooth.org/
>
> Opinions expressed herein are those of the author and do not necessarily
> reflect those of his employer.
>
>

Received on Tuesday, 10 May 2011 15:19:53 UTC