[annotator-dev] Image annotation

Sat Oct 11 13:29:47 UTC 2014

On 2014-10-11 14:50, Kristof Csillag wrote:
> Dear all,
>
> Here is a very short summary of what we have done in H's form of
> Annotator.
>
>  * First, we have moved from the current annotation data format to one
> closer to the OA model: instead of /ranges/, we have /targets//, /and
> those /targets/ are described by /selectors.
>  * /We have defined a bunch of selectors for text (again, the same
> ones used in OA), and defined one (or two) selectors for describing
> image fragments, as described by Annotorious
>  * Extended the annotation anchoring mechanism, so that the available
> anchoring algorithms can be shipped in plugins, and registered with
> the code. When a target needs to be anchored, all the registered
> anchoring algorithms are asked if they can handle it.
>  * Also, we extended the system to (logically) separate anchors (as
> the abstract information about where an annotation goes) from the
> highlight (as the actual "physical" manifestation of an anchor,
> visible to the user).
>  * We have extended the system to support different types of anchors
> and different types of highlights. (And we have implemented the
> corresponding types for image.)
>  * We created an image annotation plugin for Annotator, which bridged
> Annotorious with the framework in Annotator core.
>
> Actually, most of this work is independent of Image annotations, but
> considered in necessary, in order to do it the right way.
>
> In the end, we never really deployed the image annotation code,
> because Annotorious has a nasty habit of changing the DOM in various
> haphazard ways, and one of our other libraries (d-t-m, used for fuzzy
> text maching) was not quite up to the task of dealing with this, but
> we have a fix for that in the pipe, too.
>
> Now, this is all on our Annotator fork, and we plan to port it back to
> upstream Annotator 2.0.
>
> The latest plan to do so looks like this:
>  * We will release the framework (for handling multiple types of
> selectors, algorithms, anchor types, highlights, etc) as a separate
> library, with glue code for various versions of Annotator (A1.2.x,
> A2.0, our fork)

The /very/ incomplete and unfinished, current draft of this library is
available here:
    https://github.com/hypothesis/dom-anchor-core

You can see the readme file for a more detailed explanation of some of
the concepts, terms and procedures used by us.
    https://github.com/hypothesis/dom-anchor-core/blob/master/README.md


   Kristof


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.okfn.org/pipermail/annotator-dev/attachments/20141011/4d1830f0/attachment-0004.html>