Re: "Annotation" and "annotation" (was: RE: [data-model] Proposed Abstract for Web Annotation Data Model Spec)

Hi, Ray, Rob–

On 11/13/14 6:03 PM, Denenberg, Ray wrote:
> Rob, honestly, I don’t care much (or at all) about the abstract,

You keep saying that... but are you sure you don't care? :P

I really do care about the abstract, and about how the First Public
Working Draft is structured and presented. Web developers and
implementers will get their first impression of the Web Annotation WG
from this spec, and their interest may either be piqued or dulled by it.

I want them to look on it as a potential implementation goal.

That said, I think Paolo and Rob understand that goal well, and I trust 
them to come up with a definition that's accurate, clear, and simple.


> my comments are aimed at making the model both internally consistent
> and intuitively clear (which may be conflicting aims).

I agree with your goal, and don't think those are conflicting aims. The
data model spec should be intuitive by design, so that implementers
aren't surprised or confused, and don't get it wrong, and internal 
consistency should follow from that.


> If there is intended to be a meaningful distinction between
> “Annotation” and “annotation” – and I think it is a good idea –

This truly confuses me. Why would that be a good idea? If our aim is to 
produce an intuitive data model, why would we make distinctions that are 
fuzzy and counterintuitive?

I quoted several dictionary definitions for "annotation" [1]; none of 
them made the distinction between the content (the annotation body) and 
the description of the relationship (what Rob and you call the 
"Annotation"). In fact, they all referred to the annotation body as the 
annotation; the attachment aspect was mostly secondary (though important).

If you look at a picture of a traditional paper annotation [2], the 
thing that the average person would identify as the annotation is the 
marginalia content itself, or maybe the marginalia and the underline or 
arrow (the "link") that points to the indicated passage that that 
annotation is about.

Doesn't it make more sense to give the digital analog of an annotation 
as much verisimilitude to a traditional annotation as possible, so 
people understand it immediately?


If we really need a term for "the part of an annotation that makes an 
assertion about the relationship between the body and the target", why 
don't we coin a new term ("Annotation Assertion"?), rather than 
overloading the term "Annotation" with a conflicting or narrower 
definition? (This is bit like the problem with "punning", I guess...)


Looking at a sample annotation from the spec ...

   {
     "@id": "http://example.org/anno1",
     "@type":"oa:Annotation",
     "body": "content",
     "target": "http://example.org/target1"
   }

(or if you don't like body literals:)

   {
     "@id": "http://example.org/anno1",
     "@type":"oa:Annotation",
     "body": {
       "value": "content"
     },
     "target": "http://example.org/target1"
   }

... it certainly looks to me like the body, the body's content, the 
target, and the indicator that the object is an annotation are all bound 
up in a single package, not separate resources. That's what I think of 
as an annotation (or Annotation).


> it does not seem to be made clear in the document, and perhaps it
> should, and if so, that  would mean checking all instances of both
> and making sure they are used correctly, which I know would be a lot
>  of work.   But for example,  the first picture (which I cited) uses
>  “annotation” when (if there really is a distinction intended)  it
> should be “Annotation”.

I'd happily volunteer to create new diagrams, if we could get agreement 
in this WG to call an annotation an Annotation.


[1] http://lists.w3.org/Archives/Public/public-annotation/2014Nov/0039.html
[2] http://dypis.wikispaces.com/file/view/DiggingAnnotated0001.jpg

Regards-
-Doug

> Ray
>
> From: Robert Sanderson [mailto:azaroth42@gmail.com] Sent: Thursday,
> November 13, 2014 5:47 PM To: Doug Schepers Cc: Denenberg, Ray; W3C
> Public Annotation List Subject: Re: [data-model] Proposed Abstract
> for Web Annotation Data Model Spec
>
>
> I think the distinction is between "contains" and "references", both
>  of which are possible but have different semantics and are derived
> from different and valid mindsets.
>
> The Annotation resource in the data model is _not_ the Body
> resource... so an instance of the Annotation class in the model is
> the link between the Body/Bodies and the Target/Targets, whereas
> common parlance would have the annotation [lower case a] also
> semantically include a (typically textual) comment.  I think that Ray
> is meaning Annotation and Doug is meaning annotation.
>
> From the principles: An Annotation is a resource that represents the
>  link between resources, or a selection within a resource.
>
> Formally, the Annotation is only the aboutness, not the X and not the
> Y.  So from the model's perspective, the Annotation does not include
> X (which indeed may not even exist in the case of highlights or
> bookmarks).
>
> In the serialization, however, we return more information than just
> the Annotation resource, including information about the body,
> target, selectors, styles, agents, and so forth that any client will
>  actually need.  Therefore thinking from a document perspective,
> textual bodies are contained within the document, so it's natural to
>  conclude from this point of view that the body is part of, rather
> than referenced from, the annotation [lower case a intentional].
>
> I hope that makes sense?
>
> To move forwards, I propose that the editors take this discussion
> into account, but that we not delay FPWD for WG level word-smithing
> on the abstract.
>
> Rob
>
>
> On Thu, Nov 13, 2014 at 1:30 PM, Doug Schepers
> <schepers@w3.org<mailto:schepers@w3.org>> wrote: Hi, Ray–
>
> On 11/13/14 10:38 AM, Denenberg, Ray wrote:
>
> I don't want to get too hung up on the first sentence of an
> abstract.
>
> However, comparing;
>
> · Mine: “An Annotation asserts information about a resource”
>
> · Yours: "An annotation is a piece of information attached to a
> document or other resource"
>
> I like yours as much as mine, maybe better …. except for the fact
> that it isn’t accurate.
>
> If X is “about”  Y, X is not the annotation.  The annotation is a
> (third) resource which asserts that X is about Y.
>
> I don't agree.
>
> I know other people who agree with your concept of what an annotation
> is, but I don't think that's a useful level of abstraction.
>
> To me, and I suspect to most other people, the thing that
> distinguishes an annotation from a primary resources is that it
> contains not only content, but the link that asserts that that
> content pertains to another resource.
>
> In other words, it is both the vehicle and the payload.
>
> I think this is borne out in the data model. An annotation contains
> one or more target links and selectors, and one (zero?) or more
> bodies.
>
> Obviously, you can make an annotation that simply links two targets
> without making an explicit statement about them or their relation,
> but that's the degenerate (and less common) case.
>
> So, I'd suggest that if X is “about”  Y, (X + the "about" assertion)
>  is the annotation.
>
> What do others think?
>
> How to capture that in the first sentence of an abstract without
> blowing the  mind of a someone reading the abstract just trying to
> decide whether annotations are of interest, is admittedly difficult.
>  But I think, while the two are probably equally helpful, mine is
> more accurate.
>
> Respectfully, I think yours definition is reasonably accurate, but
> abstruse; it would be difficult for the average reader who's not
> versed in the jargon of semweb (or similar disciplines) to unpack.
>
> I don't really care about my definition per se; I do care about the
> abstract being both accurate and in plain English.
>
> Regards- -Doug
>
>
>
> -- Rob Sanderson Technology Collaboration Facilitator Digital Library
> Systems and Services Stanford, CA 94305
>

Received on Friday, 14 November 2014 06:01:23 UTC