[Clipart] developing an xml schema standard for packaged multimedia

[Nathan Eady]

Mike Traum wrote:
> I think an xml schema should be developed for a bundle of multimedia
> files. There may be one already, but in my brief search I was not
> able to come up with anything.

Maybe I'm just old-fashioned, but I would think that an archive format,
such as PKWare ZIP format, would make more sense than XML for a bundle
of files.  Even OpenOffice, which uses XML extensively, uses ZIP for
packaging up groups of related files together.  Seems to work very
well.  Java and Mozilla also use the ZIP format for similar purposes
(jar and xpi, respectively).

> This would allow metadata such as keywords and categories to be
> distributed with multimedia, allowing easier use on the client side.

Metadata, such as keywords, can be embedded into the files (if they
are in an XML-based format, such as SVG) using RDF.  For files that
are not in an XML-based format, such as WMF, we have for lack of
a better solution been placing foo.rdf right alongside of foo.wmf
or whatever.

> For example, you could then download a bunch of clipart from
> openclipart.org as an xml file. This xml file could then be used in a
> multimedia organizer application which would give you the ability to
> search keywords and see the categories.

The multimedia organizer application would need to have knowledge
of the format in question.  If we accept that limitation, though,
ZIP works just as well:

For example, you could then download a bunch of clipart from
openclipart.org as a zip file.  This zip file could then be used in a
multimedia organizer application which would give you the ability to
search keywords and see the categories.

I don't see how using XML for the container format adds anything.

> I realize that svg has the ability to have embedded metadata, but
> many other format's do not.

If you don't like placing the metadata alongside in an rdf file
within the package, another option would be to put the metadata in
a manifest file, which could be included in the package.  The file
format for the manifest file could be XML, but ZIP could still be
used for the container format.

There's a significant advantage to using ZIP for the container:
users who don't have the multimedia organizer application installed
(e.g., because they haven't bothered to install it, or because it
does not yet exist currently) can use standard tools (info-zip,
pkunzip, Nautilus, Winzip, recent versions of Windows Explorer,
or whatever) to get at the clipart.  With an XML-based container
format, that becomes impossible, and the archive is useless to
anyone who doesn't have the special application.

I would think we should use XML for the metadata, but use
ZIP or something like it for bundling up the files together.

It should also be noted that we really need to package up
thumbnails, in addition to text-like metadata such as author
and title and keywords, with each file.