Debtags Library

[Enrico Zini]

On Wed, Jul 21, 2004 at 05:08:39PM +0200, Erich Schubert wrote:

> >  - *NEW* The organization taking care of the namespace;
> Well, we should probably use the name "namespace" for these then,
> and try to keep them similar to XML namespaces.

Sounds great!  That went through my mind, then I left it in Scotland:
I'm glad you brought it back :)


> > Then you raise another issue: do we want multilevel facet/tag
> > hierarchies?
> I do want. ;-) For example for file formats.

My other proposal was: why don't we tag facets instead?  We're all here
to do something better than static hierarchies, we could just apply our
very same method to our own categorization problems :)

I mean, instead of doing this:

  tech::format::file::ogg
  tech::format::stream::ogg
  tech::format::encoding::vorbis
  tech::protocol::http

We do this:

  fileformat::ogg
  streamformat::ogg
  dataformat::vorbis
  netprotocol::http

and this:

  fileformat: tech, format, file
  streamformat: tech, format, stream
  dataformat: tech, format, encoding
  netprotocol: tech, network, protocol

Then we split the problem into a Package Browser, that looks for
packages, and a Facet browser that looks for facets.  Then you say
"Don't bother about any 'technology' tags: just give me the 'usage'
ones"...

I had this idea while writing the previous message, and I'm quite
excited by it: is there something wrong in it that I'm not seeing?


> Thinking more about the data we have and we do work with, maybe we
> should go all the way down to RDF.
> If we could keep all the stuff so generic it can work with any RDF we
> might be able to do a crazy amount of things (like browsing blogs using
> facets... ;-) )

YES YES YES!
Another project I'm about to start (yuck! too many projects!) is a file
storage with metadata, and medatata of course use tags.  Then one can
build a CMS out of it, and the website structure could lay itself out
automatically giving the file tags and file types.

Another idea I had was to make an Internet Tag Archive, and when people
browse the web they tag the pages, the tag data gets merged and there's
a unique super power bookmark website accessible for everyone.

I WANT LIBTAGCOLL EVERYWHERE!!

Only one problem: I don't know much (aka: nothing) about RDF, but
libtagcoll can handle collections of pretty much anything.  It'd be nice
to make a libtagcoll-rdf that adds a libdebtags-like layer on top of
libtagcoll with the functions to make it easy to mess with RDF.

Someone knows about RDF communities which would be interested in
something like that?  I'd be happy to help them getting started and
going with the project...


> Enrico, a couple of emails herv? and i sent you CC: yesterday bounced.
> I hope your post box fixed by now? ;-)

Yes... I was in vacation and there's been a message burst (hint:
debian-devel) that sent me severely overquota... could you and He'rve'
please resend me the mails?


Ciao,

Enrico