W3C home > Mailing lists > Public > public-openannotation@w3.org > December 2012

Re: Point vs Range/Area Selectors?

From: Robert Casties <casties@mpiwg-berlin.mpg.de>
Date: Tue, 11 Dec 2012 07:32:13 -0500
Message-ID: <50C727CD.4040500@mpiwg-berlin.mpg.de>
To: public-openannotation@w3.org
On 10.12.12 23:07, Bob Morris wrote:
> On Mon, Dec 10, 2012 at 11:00 PM, Robert Sanderson <azaroth42@gmail.com> wrote:
>>
>> Any sort of document that has geography, either linear or area.  Which, as
>> far as I can tell, is any sort of document :)
>> I think that the SvgSelector or multiple SvgSelectors will handle arbitrary
>> polygons, polylines and so forth.
> 
> Yeah, maybe. I suppose that the issues, if there are any, about
> defining polygons on raster images using vector representations, are
> not of concern to OA, but rather of the knowledge domain represented
> in the Target and Body....

+1

I don't think that we need anything special in OA for point-like
annotations (but maybe my imagination is too limited :-)

For client software that is rendering annotations on images drawing
something point-like is of course different from drawing something
rectangle-shaped or polygon-shaped.

On the annotator-dev mailing list we discussed a common JSON
representation for image annotations between YUMA/Annotorious[1] and
digilib+Annotator[2] that uses different shape types.

But each of this types would fit into one of the existing OA Selector
types so the server can map between OA and the JSON format.

Our Annotation Server[3] currently does rectangles and points as
xywh-fragments internally where points have a width and height of 0. It
does not work with OA internally but we plan to provide an OA-RDF
interface with a least read-only access. We want to do polygons soon but
we have not decided on the internal and external representation.

Personally I'd prefer to use WKT for polygons to SVG because I don't
like to have to parse XML in RDF but maybe that's just me ;-) The folks
from Maphub seem to use both[4].

Oh, and one more thing from a scaling image server developer: please
dont't use pixel units in your image annotations because they become
meaningless when the image size is lost or the size changes! Please use
normalized fractions (float 0-1) or (float!) percent.

Cheers
	Robert

[1]: https://github.com/annotorious/annotorious/wiki/JavaScript-API
[2]:
https://it-dev.mpiwg-berlin.mpg.de/tracs/Annotations/wiki/json-annotation-format
[3]:
https://it-dev.mpiwg-berlin.mpg.de/tracs/Annotations/wiki/AnnotationManagerN4J
[4]: http://maphub.github.com/api/#comment_example


>> It seems that the consensus, at least to date, is that point based selectors
>> are important, and that we don't need to do anything special for them other
>> than describe how they are modeled using range/area selectors.
>>
>> R
>>
>>
>>
>> On Mon, Dec 10, 2012 at 8:45 PM, Bob Morris <morris.bob@gmail.com> wrote:
>>>
>>> Are we talking only about text-based documents here, or in general
>>> resources that have some kind of "geography"? For example, modern
>>> digital representation of maps support-and need- selectors that are
>>> unions of polygons, and there are widely used standards for their
>>> specification. See http://www.opengeospatial.org/standards/is.
>>> JPEG2000 can encode Regions of Interest (ROIs) of arbitrary polygons
>>> and ellipses.   In general many image serializations and their
>>> application support complex ROIs.  I suppose there is little interest
>>> in annotations of medical images that could not support only
>>> rectangular selectors.
>>>
>>> Bob Morris
>>>
>>>
>>> On Mon, Dec 10, 2012 at 3:54 PM, Robert Sanderson <azaroth42@gmail.com>
>>> wrote:
>>>>
>>>> Dear all,
>>>>
>>>> Recently in a matrix of functionalities provided by different annotation
>>>> systems, I saw an entry:
>>>> * Point Based Selection
>>>>
>>>> This was distinguished as a single point in an image (eg x,y but 0 w and
>>>> h)
>>>> or a single position in a text stream (eg a selector with an offset, but
>>>> 0
>>>> length)
>>>>
>>>> Is it important in the Open Annotation model to have Point selectors, or
>>>> should we have a recommendation in the specification to use the
>>>> range/area
>>>> selectors with 0 height/width/length?
>>>>
>>>> Thanks for your thoughts on this!
>>>>
>>>> Rob
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>> --
>>> Robert A. Morris
>>>
>>> Emeritus Professor  of Computer Science
>>> UMASS-Boston
>>> 100 Morrissey Blvd
>>> Boston, MA 02125-3390
>>>
>>> IT Staff
>>> Filtered Push Project
>>> Harvard University Herbaria
>>> Harvard University
>>>
>>> email: morris.bob@gmail.com
>>> web: http://efg.cs.umb.edu/
>>> web: http://etaxonomy.org/mw/FilteredPush
>>> http://www.cs.umb.edu/~ram
>>> ===
>>> The content of this communication is made entirely on my
>>> own behalf and in no way should be deemed to express
>>> official positions of The University of Massachusetts at Boston or
>>> Harvard University.
>>
>>
> 
> 
> 


-- 
Dr. Robert Casties -- Information Technology Group
Max Planck Institute for the History of Science
Boltzmannstr. 22, D-14195 Berlin
Tel: +49/30/22667-342
Received on Tuesday, 11 December 2012 12:32:51 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 11 December 2012 12:32:52 GMT