RE: FYI: dear adobe here's an idea for you

>> I'm happy to help provide advice, guidance, etc.. Adobe is next door to
>> me, so if we've got the right contacts there, I'm happy to offer going
>> over there to present an introduction to RDFa.

The groups working on metadata for media in Adobe are spread
across the world, including Noida, Hamburg, Bucharest, Seattle,
San Francisco, San Jose and probably 8 other locations.
If you mean San Jose, I'm happy to meet and invite local folks, 
let me know.

> I'd suggest an RDFa representation of their XMP metadata format is more 
> likely to engage attention at Adobe than generic RDF/DC. 

I think what generally engages the attention of Adobe product
groups are use cases and examples that significantly impact the
work of customers engaged in creating and updating web content.
Focus on the workflow uses of metadata -- where does it come from,
and how is it used -- let the standards follow.
Of course, compatibility with existing deployed tools that support
XMP (whether from Adobe or other software sources) is a plus.
How does RDFa substantially improve things for users, authors,
publishers, over other solutions?

> Are http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart1.pdf and 
> http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart2.pdf  the best 
> places to start? 

Part 1 describes the data model (not arbitrary triples!) and the
representation of the data model within a (subset of) RDF.
This is of interest if you're transforming between RDFa and XMP.

Part 2 describes the standard schemas for XMP. This is important
if you're establishing standard vocabularies for metadata
with a shared data model.

Part 3 describes how XMP metadata can be embedded directly into
some well-known file types. One important way of associating
metadata with files is to embed the metadata -- itself in a
consistent representation -- in arbitrary file types, to allow
content-type independent manipulation of metadata. 

I'm not sure that's a goal of the "RDF in XHTML task force"
but I think it might be -- the metadata for the JPEG file
and the metadata for the HTML file that references the JPEG
should be equally processable.

Additional documents:

http://www.adobe.com/devnet/xmp/pdfs/DynamicMediaXMPPartnerGuide.pdf
covers a general mechanism for dealing with compound
media (objects derived from many parts) as well as
metadata for "temporal" media (audio, video) which
varies over time. Since this is more focused on 
"how to" it also has a more examples.

http://www.pdflib.com/developer/xmp-metadata/ has some pointers
to use of XMP in the ISO standards for PDF (now owned by ISO
and not Adobe.)

http://www.metadataworkinggroup.org/
has a specification about metadata interoperability.

For a "starting point", though, there's nothing like
running code:

http://www.adobe.com/devnet/xmp/ offers an
open source  (BSD license) C++ XMP library
implementing XMP extraction, parsing, and
manipulation.

http://labs.adobe.com/technologies/xmplibrary/
has an XMP library written in ActionScript, also
open source.


> I remember XMP used to use a lot of the rdf:Bag, 
> rdf:Seq, rdf:Alt constructs. 

I'm not sure those are used "a lot", but Bag, Seq and Alt are
part of the data model, as are structures.

> So one thing to investigate here would be 
> the extent to which data from other XMP-carrying environments (eg. 
> inside PDFs, TIFFs) can be losslessly represented in human-facing RDFa. 

I'm not sure I or Adobe would go along with the characterization
of PDF, TIFF or other formats that contain XMP metadata
as less "human-facing".

But lossless transformation between metadata representations
is an important goal for improving customer workflows and
something we support.

> And on whether XMP-friendly profiles of RDFa usage (eg. some templates - 
> such as might be suggested for Dreamweaver) can be produced.

> Here is an example from XMPSecificationPart2.pdf
> (one might imagine this in RDFa with some CSS styling):

"<xmp:Title>
<rdf:Alt>
<rdf:li xml:lang="x-default">XMP - Extensible Metadata Platform</rdf:li>
<rdf:li xml:lang="en-us">XMP - Extensible Metadata Platform</rdf:li>
<rdf:li xml:lang="fr-fr">XMP - Une Platforme Extensible pour les
Métadonnées</rdf:li>
<rdf:li xml:lang="it-it">XMP - Piattaforma Estendibile di Metadata</rdf:li>
</rdf:Alt>
</xmp:Title> "

The example confused me, since "dc:title" is in the Dublin
Core namespace (http://purl.org/dc/elements/1.1/), not the 
"xmp" namespace (http://ns.adobe.com/xap/1.0/). 

But I think s/xmp:Title/dc:title/  would help.

> Larry - looking at the spec, a lot of the examples are exercises in 
> exploring corner-cases for parsing. Can you point us to any kind of 
> repository of real world XMP examples?

I'll look to see if anyone has published or could publish some
XMP examples to help people working on interoperability.

Thanks!

Larry
-- 
http://larry.masinter.net

Received on Tuesday, 24 February 2009 21:23:40 UTC