On Wed, 23 Apr 1997, Stu Weibel wrote:
> Andrew Daviel writes:
>
> > We need a consensus on how to specify metadata for non-HTML Web objects,
> > IMO.
>
> The infrastructure for supporting metadata of non-html Web objects is
> not now in place, but will be within the year. It is PICS-ng.
Is this http://www.research.att.com/~presnick/pics/0196wg/labchart.htm
(Next Generation Label Format) ??
Compared to Dublin Core, PICS seems complex and geared towards filtering
and real-numbered values, at least in the minds of consumers. The
transport mechanism is, as I understand things:
- by META HTTP-EQUIV=" PICS-Label"
- by HTTP PICS-Label:
- by 3rd-party ratings bureau.
If used as an HTTP header on a non-HTML object, it has bulked out
the header. If from a ratings bureau, there is no defined mechanism
to refer to the metadata from the object itself (one could
perhaps use Link: <url>;rev="PICS-Label" ??)
Within the framework of my original suggestion, I guess one could say:
<html><head><title>Apple Blossoms</title>
<meta http-equiv="PICS-Label" content='(PICS-1.1
"http://purl.org/metadata/dublin_core_elements/PICS"
by "Ann Photog" for "http://www.some.org/blossom.gif"
ratings (dc.title "Apple Blossoms" dc.creator "Ann Photog"))'>
</head><body>
(<a href="blossom.gif">here's the photo</a>)
</body></html>
etc.
I believe that some PICS ratings bureaus work by sending a robot to cehck out
the URL in question for flesh-toned GIFs and naughty words, with a later
editorial step; a discovery "ratings bureau" might operate in the same way,
so we still need a mechanism which is easy for authors to understand in
order to associate original metadata with non-metadata-capable objects
Andrew Daviel
|