Re: Non SemWeb uses of RDF

Hi Brian,

* Brian Manley <manleyr@telcordia.com> [2004-12-17 17:44-0500]
> 
> Hi Frank,
> 
> "how do you define a 'non semantic web application'"
> 
> This is a hard question. I think I know a semweb app when I see it, 
> but I'm not sure how to define it. I think ( perhaps naively ) 
> of semweb applications as those that either produce or consume RDF 
> metadata relating to web-accessible resources. RSS aggregators, 
> browsers, etc. fall into this category. On the other hand, a non 
> semweb application doesn't really concern itself with whether or 
> not the resource it's describing is a web-accessible resource or 
> not. It could be a web page about rocks, or an actual, physical rock. 

I don't find this a helpful way to draw the distinction. Sorry!

Guess that means I need to sketch an alternative. Hmm. Let me try.

The phrase "Semantic Web" is a slogan for our efforts to make Web-based 
systems more sensitive to some aspects of the meaning of documents. In 
particular, about what those documents tell us about various kinds of 
relationships between all kinds of thing, offline and on. eg. documents are 
made by people, and describe people, places and ideas etc. All metadata 
efforts, whether labeled "Semantic Web" or not, have grappled with this. 

The Resource Description Framework is a piece of W3C technology designed to 
progress the Semantic Web project, by providing a domain-neutral set of 
conventions for describing things (whether people, documents, events,
places, whatever...) in terms of named relationships. RDF, even before
we started using the phrase "Semantic Web", is *always* happy mixing 
descriptions of online/electronic and offline/realworld things. In many 
ways, that was the point. Before RDF, metadata initiatives tended to 
get started with a particular kind of thing or topic area as their 
central focus. This was great, except it meant that we ended up with 
people working on file formats, XML/SGML tagsets etc each oriented towards
"Education" or "Images" or "Rights", or "Computer Science Reports" 
or "Software" or "Mapping/Location". RDF was created through W3C's 
Metadata Activity as an acknowledgement that these different forms of 
metadata were describing a common, rich, highly linked world, and that mixing 
them together was a recognised need in the metadata community 
(eg. see the Warwick Framework efforts associated with the Dublin Core
community). 
 
Time passed, and the slogan "Semantic Web" got deployed, in part 
because the concept of "metadata" is a tricky thing to define, and "data 
about data" doesn't quite work when you look at the details. Often your 
"data about data" turns out to be "data about real world entities
associated with your data", eg. the location of a photo, or the 
contact info for an author, or the price of something. It was also 
an acknowledgement that the RDF design was very general purpose, and 
applicable to fairly arbitrary data interchange, rather than the 
application areas traditionally associated with the label "metadata".

So, RDF is a technology, a tool, a means to an end. "Semantic Web" is 
our name for that end: a more semantic, ie. meaningful, Web. All RDF 
applications help, to some extent, in that direction, eg. since users of the 
technology (eg. by supplying feedback on specs, bug reports and 
encouragement to tool developers) help our collective efforts. If some 
RDF projects and applications are "more Semantic Webby" than others, I 
think the distinction is most usefully made in terms of thinking 
about contributions to the larger Web project, begun 15 years ago. The 
Semantic Web effort is explicitly part of this shared enterprise, and 
RDF projects which emphasise the Web (open, free exchange of data, 
collaboration, shared tools and interfaces, freedom from data lock-in,
pluralistic technology that allows diverging views to share 
infrastructure, etc.) are particularly valuable contributions to the 
Semantic Web initiative. 

> My limited googling and research seems to indicate that the semweb 
> community itself is a bit fuzzy on the question too. True? 

I think there's a little truth in that. Many in the Semantic Web 
community at the moment are particularly excited by the possibilities
raised by formal, machine-friendly representations of meaning. In 
particular, the sorts of simple rules that can be expressed in 
OWL ontologies and languages such as N3. There has in the last 
few years been a lot of attention given to the "Semantic" part of 
"Semantic Web", perhaps at the expense of the "Web" aspects. But it
all balances out over time, and RDF data is usable as data 
regardless of whether it is consumed by a fancy, inference-capable 
KR system, or a simpler data-oriented triple store. So sometimes 
"Semantic Web" can get seem to just mean "that inference and logic 
and ontology" stuff, a characterisation which has led some 
cautious, skeptical or just data-centric folks to 
characterise their efforts as "RDF but not Semantic Web". 

FWIW, I concern the FOAF project (http://www.foaf-project.org/) to 
be a Semantic Web effort, a contribution to this larger project to 
improve the Web. It uses RDF, RDFS and OWL to help make "homepage-like" 
data more suited for machine processing, and in doing so helps people 
create RDF descriptions of both people and documents, amongst other things. 
In this sense, I think it would fall under your definition of non-semweb.
But it really is intended as a practical contribution to the 
semantification of the Web... 

Not sure if this helped, but it's late so I'll stop scribbling and send.

cheers,

danbri

Received on Saturday, 18 December 2004 01:44:49 UTC