Re: Roles, multiple bodies, and creating wiki page of potential solution

Hi Tim,

That works for literals but not URIs, and was one of the deficiencies in
the CG model that we were trying to get rid of for SemanticTags.

(e.g. a Video could be a Comment and a Reply in different annotations)

Rob


On Wed, Aug 12, 2015 at 9:11 AM, Cole, Timothy W <t-cole3@illinois.edu>
wrote:

> One more semantically valid approach to raise as a compromise possibility
> for the multi-body / role issue. We currently allow this:
>
> "body": "paris"
>
> If you want to identify "paris" as a tag, then you clearly have to say a
> little more. We currently tell you to do this:
>
>   "body": { "type": ["Tag", "EmbeddedContent"],"value": "paris"}
>
>
> How unnatural is this to plain JSON developers?
>
>
> If it is acceptable, then I would propose extending this approach to do
> this:
>
>   "body": { "type": ["Comment", "EmbeddedContent"],"value": "paris"}
>
> (where Comment can be replaced by versions of our motivation vocabulary...)
>
> -Tim Cole
>
> ________________________________________
> From: Cole, Timothy W
> Sent: Monday, August 10, 2015 17:03
> To: 'Frederick Hirsch'
> Cc: 'W3C Public Annotation List'
> Subject: RE: Roles, multiple bodies, and creating wiki page of potential
> solution
>
> For those who might be interested, I did create a wiki page illustrating
> most of the proposed ways so far raised to describe (and thereby
> differentiate) the non-multiplicity roles played by individual Bodies /
> Targets included in a multi-Body / Target Annotation:
>
>
> https://www.w3.org/annotation/wiki/Expressing_Role_in_Multi-Body_Annotations
>
> I created 2 multi-body annotations and 1 multi-target annotation as
> examples (scenarios), and then created serializations in both JSON and
> Turtle illustrating the various approaches. The annotations created lack
> required annotation provenance, but otherwise include (I think) all of the
> MUST and SHOULD triples required by our data model.  The multi-target
> scenario is a bit of a stretch -- I suspect either we don't need to worry
> about multi-Target use cases or we need to revise and maybe augment our
> current list of Motivation terms to better handle multi-Target use cases.
>
> The first example includes 3 bodies: a Textual Body, a SemanticTag Body,
> and a SpecificResource derived from a YouTube Video.
>
> The second example has 2 Textual Bodies, and the third example has 2 Web
> page Targets.
>
> The approaches illustrated have mostly been discussed on the list, except
> for the last approach which would type Bodies and Targets according to
> their role in the Annotation (which will likely make Rob cringe a little).
>
> Anyway, may be helpful for our discussion Wednesday and beyond in case you
> find the email links Fredrick provide not quite enough.
>
> Rob, Fredrick, since this page is transient, I did not try to link from
> the WG Wiki's home page.  Feel free if you think that useful.
>
> Thanks,
>
> Tim Cole
>
>


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