Re: [Fwd: Re: [MMSEM] RDF and syntactic interoperability]

Dear Susanne,

So below you address a very common problem that fortunately has been quite
discussed in the litterature and several solutions have been proposed.
First, I have partially answered to your problem in [1].
A detailed answer is given below.

> the problem is the following. Imagine that I use RDF to describe
> a resource such as an image. How to I model in RDF that I
> want to refer to a fraction of this image. This might be needed
> if want to express the following.
>
> Subject (Susanne) - Predicate (is_the_person_in) - Object (rectangle
> in an image)

So, as I said, your problem is not RDF, is how to localize and uniquely
identify a region of a multimedia document (you give the analogy in the
SMIL world). That has nothing to do with RDF and is more linked to a URI
problem (see below).
This problem is slightly discussed in [2] (section 3) but extensively
addressed in [3, 4] and particularly [5]!

In short, you can imagine several tricks to solve your problem.

1/ Using SVG bounding box mechanism:
This is what PhotoStuff is doing! In other words, you define a region using
an SVG snippet, you can even add an anchor name, so the region is then
identified and your RDF annotation will be "about" this anchor. You can
find an example at [6].

2/ Using MPEG-7:
Again, you create an MPEG-7/XML file (identifiable by an URI) that contains
the definition of your region. Here, you have all the expressivity of
MPEG-7, that is, the possibility of creating non-spatially connected region
(mask in MPEG-7), typing your region (ellipse, rectangle, polygon, defined
with exotic mathematical transformation, etc.) ... and again, your (RDF)
annotation will be about the URI fragment defining this region (i.e. the
XML portion of this file).

Both solution introduces an indirection. Your annotation is not "about" the
media itself but "about" an XML file that defines a region in a media. Note
that for defining Temporal interval of video or audio document, a nice
proposal named TemporalURI is on the table and I would strongly encourage
to use it! The apache module server even exists to serve you only the video
bits corresponding to the time interval you specify in a URI (not the whole
video!). I would say this a clean solution and I'm all in favor to develop
the SpatialURI based on the same model ...

Best regards.

    Raphaël

[1] http://lists.w3.org/Archives/Public/public-xg-mmsem/2007Mar/0010.html
[2] http://www.w3.org/2005/Incubator/mmsem/XGR-image-annotation/
[3] http://ftp.cwi.nl/CWIreports/INS//INS-E0308.pdf
[4] http://ftp.cwi.nl/CWIreports/INS//INS-E0309.pdf
[5]
http://www.uni-koblenz.de/~aggrimm/arbeitsberichte/arbeitsberichte_4_2007.pdf

[6]
http://www.w3.org/2005/Incubator/mmsem/meetings/f2f-amsterdam/multimedia/annotations/jacco_try.rdf

[7] S. Pfeiffer, C. Parker, and A. Pang. Specifying time intervals in URI
queries and fragments of time-based Web resources. Network Working Group,
Internet-Draft, 2005. http://www.annodex.net/TR/URI_fragments.html.

--
Raphaël Troncy
CWI (Centre for Mathematics and Computer Science),
Kruislaan 413, 1098 SJ Amsterdam, The Netherlands
e-mail: raphael.troncy@cwi.nl & raphael.troncy@gmail.com
Tel: +31 (0)20 - 592 4093
Fax: +31 (0)20 - 592 4312
Web: http://www.cwi.nl/~troncy/

Received on Monday, 2 April 2007 09:16:56 UTC