[annotator-dev] Branching v2.0.x

[Robert Sanderson]

Hi all,

Thanks for picking up the Open Annotation in Annotator work :)

Weighing in on a particular point:

Randall wrote:

At this point, I'm convinced the model should change. For instance, it
disambiguates the meaning of tags on annotations. Right now I feel they're
ambiguous, referring either to comment or the target. If the tags were the
annotation body, rather than a field on the annotation, they would more
clearly apply to the target.


This is precisely the reason for tags being modeled as full annotations in
the Open Annotation model, and for using ContentAsText to model bodies
rather than using a string literal.  If there is identity for everything,
then everything can be annotated unambiguously.

In particular, the Open Annotation model says that all bodies of an
annotation are about the target of the annotation.  So if a single
annotation has two tags and one comment, then all three of those things are
about the target(s) of the annotation, not the annotation itself.

There are other ways to tag resources, such as simply adding fields into
the model. However the complexity becomes:
1.  Interoperability between systems is now out of the window -- most will
just ignore extraneous fields and the tags disappear
2.  Even within a system, if someone tags an annotation in the correct way,
you would have to process two different tracks -- the short cut added
field, and the full tag-as-annotation method.  So it's easier in the long
run to add in the work up front, and have an unambiguous model down the
line.

Randall wrote:

If we get the shape of the model right, contextualizing it with JSON-LD can
be done in plugins of various sorts, such as for a storage plugin that
interfaces with JSON-LD endpoints.

I'm happy to help where I can with this implementation work if there's a
design for where to put the code.  My naive thought was simply to
re-implement store.jsonify() to produce Open Annotation JSON-LD, but
there's doubtless a better way :)

Hope that helps,

Rob Sanderson
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.okfn.org/pipermail/annotator-dev/attachments/20131030/93914140/attachment-0002.html>