A proposed theme for SW Activity phase 2: 'SW on the Web'

Attached is a message I've just sent to the W3C Semantic Web Coordination Group
list, where we are discussing priorities to propose for phase two of the 
Semantic Web Activity, beginning later in the spring.  The mail archives for SWCG are 
on the W3C Member website at http://lists.w3.org/Archives/Member/w3c-semweb-cg/
so passwords are needed to track that discussion.

Needless to say the attached view[1] is just my personal opinion. I'd be interested 
in the views of SWAD-Europe folk on how they see things, personal views or otherwise.

More context: W3C SW page, http://www.w3.org/2001/sw/ and current 
Activity statement: http://www.w3.org/2001/sw/Activity#

Dan


[1] http://lists.w3.org/Archives/Member/w3c-semweb-cg/2003Jan/0032.html

attached mail follows:

Short version: phase two of W3C SW Activity should focus on getting RDF used in the 
public Web. Everything else (especially new technology creation) is secondary to that 
goal.


Here's a longer account of my view for SW Activity phase two priorities.

(Ontologists, note that I write 'RDF' in its broader sense, as a Framework, for which 
we now have an Ontology language; when I write 'RDF' pls read 'RDF/OWL' if you prefer).

RDF and SW/Ontology tools are very general. They could be used for almost anything 
to do with modern information management. This broad applicability is what 
draws many of us to the technology, but also a source of risk: by being 
_usable_ for everything, we risk being _used_ for nothing. A reasurringly wide 
range of groups are exploring the possibilities of RDF, yet RDF is not tailored to 
any of their specific needs. We must be doing something right. But we shouldn't forget 
that more tailored solutions in each field could easily discourage RDF adoption.
We can't please all the people, all the time.

Something we might attempt in phase two is to focus our effects a bit on some 
particular fields for deploying RDF/OWL. I propose (this is so obvious as to 
barely need saying, perhaps) that we focus on the public Web in phase two of the Activity.
By 'the public Web', I mean the work of Web masters, HTML authors, focussing on 
publically available sites who are currently mostly using HTML+CSS+jpeg/gif/png as their 
content formats, and perhaps dabbling with RSS, Flash, SVG and other fancy new stuff.

I would like to see more RDF files on publically visible Web servers. RDF files that 
use a variety of RDF schemas and Ontologies, and that link to other RDF documents 
scattered around the Web. Once we have this, I'm confident the rest (intranets, domain
specific tools, RDF in backend databases etc.) will follow. Right now, if you go 
looking in the public Web for folk using RDF, there really isn't much out there. A few 
large datadumps, a few keys into webservice lookups or screen scrapes, a fair amount 
of Dublin Core embedded (invalidly) in HTML or linked as separate files, a promising 
number of RSS files, some of which are in RSS 1.0 (but not really exploiting the RDF 
aspects of RSS), and a few hundred FOAF files. This is pretty modest situation to be 
in after 5+ years of RDF work. It isn't disasterous, but should be a cause for concern, 
and for focus as we design the next phase of the Activity. 

I would like us to go into phase two with some notion of success criteria: what would 
count as having succeeded? I'm suggesting that an almost quantitative approach 
may be applicable. If there are lots of RDF documents being used on ordinary Web sites,
whether commercial, personal, Weblog, portal or academic, then something is going 
right. If we end phase two without this, then we likely should think about packing 
up and going home. 

Let's make phase two all about roll-out. Getting RDF, even simple, perhaps boring RDF, 
on the agendas, web sites and CVs of ordinary Web masters and content producers. That 
RDF will be all the more meaningful if it draws on vocabularies enriched with OWL, and 
we will have our work cut out for us helping folk do this effectively.

If, in attempting to get RDF deployed in this way, we find the reason for resistence 
is that people need common APIs, more standard query languages, protocols, or rule 
languages, then yes, we could start work in those areas. But right now I don't believe 
for a second that RDF is relatively undeployed because we've not created enough 
accompanying Web standards. When RDF is used, eg. in RSS, or Dublin Core, it is often 
'on faith', ie because people are anticipating some payback from using it where 
they could have adopted a pre-RDF or vanilla XML solution. Many people are still 
waiting to see what their RDF dabblings bought them.

By focussing phase two on public Web deployment, I believe we will have a good 
chance of increasing the number of RDF documents and RDF tools and RDF-basd services 
that are visible to folk investigating the technology. Since RDF is all about 
data merging and network effects, it becomes a stronger platform for information 
management with every new shared document that uses it. This, if nothing else, should 
draw our attention towards priorities that encourage the publication of RDF/XML 
documents on the Web. 

If this account of a theme for phase two is at all persuasive, I think we could 
derrive some specific work items and priorities. But let's agree on what we're trying 
to achieve, first. Even if it something as almost-crass as 'lots of RDF documents 
on lots of Web sites'. Stating the obvious might be worthwhile, I suspect.

cheers,

Dan


ps. i'll probably forward this to a public archived mailing list, for 
b/g in discussions in SWAD-Europe and the RDFIG.