::scr semantic

Andy Wardley scr@thegestalt.org
Mon, 5 Aug 2002 12:28:37 +0100


On Mon, Aug 05, 2002 at 11:31:50AM +0100, Struan Donald wrote:
> Heck neither have i really but... All this RSS stuff seems not at all
> like what I take the semantic web thing to be. It all just seems like
> another delivery mechanism for the sam content. It's a delivery
> mechanism that goes and gets the content for you rather than you
> having to check for it but that's all really.

I agree that RSS doesn't live up to the semantic web promise (but then
what does?), but I think that's largely a factor of the way in which
most people use it, rather than a flaw in the original concept.

And in the specific case of XML, RSS, RDF, etc., the concept rocks, but 
the implementation sucks :-(

> If it was the semantic web then it'd not only get the content for you
> but understand something about the relationship between the content
> it's getting and other content out there. 

As I understand it, the fundamental aspect of the "semantic" part of
the "semantic web" (or indeed "semantic $x") is to provide a mapping
between the "thingy" that some content is marked up to describe, and 
a corresponding "thingy" that has some common, everyday, human meaning.

Vanilla XML implements a "syntactic web".  You can parse a document,
see a tag like "<title>Lord of the Rings</title>" and "understand" that 
"Lord of the Rings" is the "title".

But because this mapping is purely syntactic, you simply know that it's 
identified by the key "title" and you don't know if it relates to the 
title of a book, or the title (honorific) of a person.  They're human 
concepts that have a higher level meaning.

That's where the semantic part comes in.  RDF allows you to assign
meaning to that markup.  In effect you can say, "this title here is
the title of a book, for the meanings of 'title' and 'book' as defined
in the Dublin Core Hand-Waving document Version Mumble.Mumble".

The obvious application is in a search engine.  You, as a human, can
identify the fact that you're looking for the title of a book and 
indicate it in some way to get a better search result. 

There are many other examples and evangelism about how this is such
a super cool thing.  The links that Jo posted cover this in great 
detail.

> Much like a librarian will know that if you are looking
> in the neurochemistry section then there may also be relevant
> information in the biology, psychology and chemistry sections. 

And there's the rub.  Things like RDF can be useful, if used properly,
to make some hard things easier.  But it's no silver bullet.  
While we can use it to make connections between markup tags and 
some level of human meaning, we can't use it to automagically implement 
higher level human understanding.

> RSS is, as you say, just posh screen scraping largely. Or at least it
> as as used at the moment. 

Yep, what he said.

A