RDF People Crawler (Photographic)
Tue, 14 May 2002 11:35:34 -0400
I would point your friend and various other efforts to embed
data in a JPEG. One of these is Adobe's XMP initiative which
is pretty open and, using RDF, allows information from extension
vocabularies to be handled very easily.
From the architectural point of view, the Scandemonium design
suffers from a centralized lookup at Scandemonium.com.
Using a guid instead of an http: URI puts the onus on
that company to preserve the lookup table, rather than the
whole community using distributed DNS and HTTP servers.
So, while it may draw a lot of traffic to their site, it isn't a
scalable general solution.
XMP, however, does seem to be.
On Sunday, May 12, 2002, at 10:12 PM, Jeff Barr wrote:
> A friend of mine has been working on a way to embed a
> pointer to Metadata in an application header of a JPG file.
> Here's his site:
> ----- Original Message -----
> From: "Rohit Khare" <Rohit@knownow.com>
> To: <firstname.lastname@example.org>
> Cc: "Adam Rifkin" <email@example.com>; <firstname.lastname@example.org>;
> <email@example.com>; <firstname.lastname@example.org>
> Sent: Friday, May 10, 2002 5:52 PM
> Subject: RDF People Crawler (Photographic)
>> So DanBri mentioned a fascinating idea/project he's been working on at
>> W3C that may well be the first entertaining application of RDF I've
>> heard of.
>> IF you take the effort to mark up your digital camera snapshots with
>> who's who
>> THEN we can automatically generate a sequence of photographs chaining
>> you to anyone else
>> Six-degrees-of-separation crossed with natural exhibitionism should
>> yield a very pretty brew of odd outcomes. I wonder, for example, what
>> chain of snaps could possibly lead from me to Kate Hudson. Or, for that
>> matter, Ekta Kapoor :-)
>> That is to say, the pleasure button is the shutter button.
>> See, it's too complicated to violate privacy of buddy lists; it's too
>> cumbersome to ask people to decorate their address books; it's too
>> inaccurate to just google for strings. But simply labeling photos with
>> who, what, when, or where, is a reasonably useful thing to build into
>> tools like iPhoto or weblogs. DanBri pointed out that 'co-depiction' is
>> a helpfully neutral lowest-common-denominator.
>> I also wonder if HotOrNot is a possible bootstrap database. Certainly
>> know Yahoo!'s most-emailed photo service gives us a bunch more links.
>> And even existing 'album sites' have some hope of being autoscannable
>> AI readers like SBook.