W3C home > Mailing lists > Public > public-openannotation@w3.org > February 2013

Re: body which is a non-information resource

From: Robert Sanderson <azaroth42@gmail.com>
Date: Mon, 25 Feb 2013 16:33:59 -0700
Message-ID: <CABevsUHvnp+GS-RY5iwk2QGBYDA0rJ=DsQadCFeFNPb-3Wsr0g@mail.gmail.com>
To: Steve Baskauf <steve.baskauf@vanderbilt.edu>
Cc: public-openannotation@w3.org
Hi Steve,


On Mon, Feb 25, 2013 at 8:25 AM, Steve Baskauf
<steve.baskauf@vanderbilt.edu> wrote:

> <http://guid.mvz.org/relations/23423> a dwc:ResourceRelationship,
> oa:Annotation;
>                                      oa:hasBody
> <http://arctos.database.museum/guid/MVZ:Mamm:14523>;
>                                      oa:hasTarget
> <http://arctos.database.museum/guid/MVZ:Mamm:14524>;
>                                      oa:motivatedBy
> <http://rs.tdwg.org/relations/offpringOf>;
>                                      oa:annotatedAt "2001-09-14";
>                                      oa:annotatedBy
> <http://guid.mvz.org/agents/James_L_Patton>.
>
> <http://arctos.database.museum/guid/MVZ:Mamm:14523> a
> dcmitype:PhysicalObject, dwctype:PreservedSpecimen.
> <http://arctos.database.museum/guid/MVZ:Mamm:14524> a
> dcmitype:PhysicalObject, dwctype:PreservedSpecimen.

Very nice! :)



> There are several issues that come to my mind with this kind of use:
>
> 1.  According to the OA model description, typically the Body "is the
> comment or other descriptive resource" and the Target is a thing that "the
> Body is somehow 'about' ".

Yes, we were careful to put in weasel words like "typically",
"generally", "frequently" and so on when trying to define what we
meant by Annotation, Body and Target.  The sort of thing that most
people think of when they hear "annotation" is a comment about
something, and we didn't want to define it such that it would be
surprising.

However we definitely recognize other use cases, such as linking,
classifying and identifying where the body is in no way "about" the
target.
What we want to avoid, however, is re-inventing RDF inside RDF where
an Annotation ends up just being a triple with subject (body),
predicate (motivation) and object (target).

> In my example the Body is a dead mammal which is
> a non-information resource and is in no way descriptive.  The Body is
> related to, but not really "about" the Target, although one of the listed
> instances of oa:Motivation is oa:linking, and asserting that one organism is
> the offspring of another is a sort of linking.

Yes, that's perfectly okay.


> 2. I'm a little confused about what one is doing when one creates an
> Annotation.  I think that creating an annotation is the act of asserting
> that there is a connection between the two resources.  However, the
> description of the oa:Motivation class says that a Motivation instance is
> the reason for the creation of the Annotation instance, NOT the reason for
> the creation the relationship which the Annotation instance documents.  Are
> the various items on the list of instances of oa:Motivation things that we
> are saying an annotating agent has done?  Or are those things that we are
> saying that the annotating agent is asserting has been done?  For example,
> if an agent creates an Annotation instance with motivation oa:commenting, do
> we assume that agent has actually created the comment or that the agent is
> just documenting that a comment has been created by someone else?  In my
> example above, the annotating agent cannot have a role in the creation of
> the relationship between the Body and Target.  The agent is simply recording
> the existence of the relationship.

That's a very good point, and it isn't clear in the documentation, I
agree.  The Motivation is information about the relationship between
the resources, and not necessarily to do with the agent that created
the Annotation.  For example, I could create an Annotation where
someone else's YouTube video is the body, and the target is an image
on the web.  I didn't do the commenting, but I'm asserting that the
reason the annotation was created was to model the fact that the body
resource is commenting on the target resource -- this seems to be
exactly equivalent with your description: The agent is simply
recording the existence of the relationship.


> Using the OA model to document dwc:ResourceRelationship instances is very
> appealing to me.  I'm just not sure if it is appropriate to use OA to
> describe relationships that may not fit the description of oa:Annotation
> instances, and with Body resources that are non-information resources.

I think it should be fine :)

Hope that helps!

Rob
Received on Monday, 25 February 2013 23:34:26 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:38:22 UTC