Re: CFC: Basic Roles Proposal

On Tue, Sep 1, 2015 at 7:04 AM, Benjamin Young <>

> On Tue, Sep 1, 2015 at 9:27 AM, Denenberg, Ray <> wrote:
>> Rob – My thoughts on 3.1 really have to do with  the question I asked
>> yesterday.  It seems to me that we do not have a clear understanding of the
>> distinction between a motivation and role.
To try and provide some spec-like text, assuming that motivations are only
allowed on SpecificResources:

A motivation is a resource that identifies the intent behind the inclusion
of the source resource in the annotation.

None of the examples in 3.1 shows a motivation and of course that’s because
>> it’s about roles.  But I think there should be examples that show both a
>> motivation and one or more roles so we can better understand the semantic
>> relationship.
Except that as per 3.2.5, we might want to remove motivation from
annotation completely.  Hence I left them off the examples.  Also the
motivation on the Annotation would just be the set of motivations on the
specific resources.

>  For example in 3.1.7 there are three roles (1) comparing (2) antecedent
>> (3)  subsequent
>> Clearly “comparing” is semantically the same as a motivation and
>> “antecedent” and “subsequent” are not.  So the annotation would have the
>> same meaning if “comparing” were to be listed as the motivation with no
>> role assigned to the body.
I couldn't come up with a gerund for antecedent and subsequent :)  But the
usage is the same -- the intent of the inclusion of the first target is
that it is the thing being compared to the second target.  It's not a great
example, I know, and would be happy to replace it with something else.

>   When this whole business came up (motivations on individual bodies) it
>> was to support the ability to, in a sense, combine a lot of annotations
>> into a single annotation, for efficiency.  I think that idea was rejected,
>> but the idea morphed into the ability for  bodies to have “roles” in
>> support of the (annotation-level) motivation.   Is this an accurate
>> characterization?
It's to provide more information on a per-body level, in the same spirit as
Motivation does for single body annotations (and Tag/SemanticTag classes do
in a hacky way for tags).

Take 3.1.1 for example.  The role is “commenting”.  That could be expressed
>> with commenting as the motivation, with no role on the body.   I’m afraid
>> that when people read this they are going to wonder what is the semantic
>> difference between these two annotations, when there is none.
Indeed! And hence I support 3.2.5 (though it should be split up to avoid
conflating the renaming with removing motivation from the annotation)

>    So I would say, for one thing, if there is only one body, it should
>> not have a role, but that role should instead be expressed as a motivation.
>> I would say further that there should be no roles expressed unless there is
>> an annotation-level motivation that the roles support.
> I think this highlights the confusion that having both annotation and body
> level motivation/roles is bound to create. It's quite likely that if we put
> these in both places, there could be conflict: an annotation marked as
> "editing" with bodies for "replying", "tagging", and "highlighting."
We'd end up with scenarios where we'll need to enforce (somehow) which
> bodies are available when the annotation uses a certain motivation--which
> trends us back towards classes of annotations...which I think has been
> dropped as an option a few times.

Yes.  But regardless, none of the above is about the changes in section
3.1.  3.1 is intentionally silent about all of the changes in 3.2, and I
tried to be as minimal in the examples as possible (causing some confusion,
apologies, re language and format) to highlight only the role aspect.


Rob Sanderson
Information Standards Advocate
Digital Library Systems and Services
Stanford, CA 94305

Received on Tuesday, 1 September 2015 15:06:58 UTC