RE: Cursors face defining moments on the Web

Someone pointed out "Third Voice" is somewhat similar to what I described,
and I also appreciate the link to "Group Lens".  Both of those services
could be implemented against "metadata servers" as I was describing.
Collaborative filtering, annotations, ratings, etc. - all of these could be
viewed simply as tools that utilized the cloud of metadata.  All of these
systems seem to store their metadata in proprietary formats, though.  Until
*anyone* can save metadata about anyone else's resources, and *anyone* can
build tools to access and manipulate the metadata, tools like this will be
to the SW as hypercard was to the WWW.

As for the "annotation server" being available on multiple platforms, why
not just define a spec describing how agents interact with a metadata server
(xml updates and queries using RDF maybe?) and then people could just
implement the servers any way they wanted.  Something that retrieved,
aggregated, and returned RDF packets would be really simple to whip up
shimmed to Oracle, SQL Server, etc. and people could use any other
underlying implementation so long as it matched the spec.  I'm thinking of
the way I can write an HTTP server using any language I choose, and I can
store web pages in any data format I choose, so long as the web pages are
HTML by the time they get sent out port 80.  By the same token, the RDF
server could listen on port 80 and return metadata about URLs.

Clearly it would be very important to keep the RDF servers separate from the
content servers, to protect the ability of individuals to link metadata to
*any* resource.  Just like the web allows an individual to hyperlink to any
page.  I think there are some things that would *not* be really a good idea
to resolve right away, such as:
* Rich query capabilities.  So long as the api allowed simple queries and
didn't prevent richer queries from coming along later
* Synchronization architecture - things like who is going to run the
servers, why, and how they keep in synch are very interesting but also not
necessary to get started.  To start the client should be able to assume that
the "cloud" will always return an answer if there is one.  Things like
redirecting or routing queries, clustering data, etc. should maybe be added
later.

In other words, I am saying, HTML is to HTTP as RDF is to X??  Why not make
a really simple definition of X (and I think X==HTTP would be a terrible
idea) and then go for it?  Put one or two in place, and let people start
making tools..

> -----Original Message-----
> From: Sean B. Palmer [mailto:sean@mysterylights.com]
> Sent: Thursday, February 01, 2001 8:45 AM
> To: Charles McCathieNevile; William Loughborough
> Cc: Al Gilman; Joshua Allen; Rob Cameron; WAI ER group; Marja-Riitta
> Koivunen
> Subject: Re: Cursors face defining moments on the Web
> 
> > annotations work being done in Amaya.
> 
> Until there is a permanent annotations server (is the W3C one permenent
> yet?), or, even better, fully ported annotations server software (I run
> IIS... why are there no tools for IIS???; I end up having to write my
own),
> I doubt it will take off. Amaya is good because it attempts to bridge the
> UI gaps that TimBL mentioned [1], but I don't find it intuitive enough to
> use at the moment for XHTML. Only plain text tools are direct enough for
> XHTML, IMHO. [The DIAP problem has been getting progressively worse now
> for
> ten years, and it'll probably never be fixed.]
> 
> [1] http://www.w3.org/DesignIssues/Editor and
> http://www.w3.org/DesignIssues/UI
> 
> --
> Kindest Regards,
> Sean B. Palmer
> @prefix : <http://webns.net/roughterms/> .
> [ :name "Sean B. Palmer" ] :hasHomepage <http://infomesh.net/sbp/> .

Received on Friday, 2 February 2001 01:03:41 UTC