Re: ::scr semantic

[Simon Wistow <simon@xxxxxxxxxxxxxx>]

On Mon, Aug 05, 2002 at 11:30:38AM +0100, Alaric Snell said:
> Hmmm... you guys are talking more about Web Services than Semantic Web, here.
> 
> Web services = stuff like RPC or using HTTP to fetch a 
> not-necessarily-human-readable file from a bit of software, like those 
> syndication things.

"And lo, the scales fell away from his yes and all was revealed unto
him" 
	Muttley 1:17


I think this is where I've been going wrong. I'd always herad of the
semantic web in thatsort of RSS/RDF/SOAP/Services context and that's why
I couldn't get it is the idea to end all ideas.

Instead, what Jo is talking about, I think is a way of encoding all our
information in a computer readable form and then aggregating and
extrapolating that data to make assumptions and statements.


Vxy likes(x,y) -> likes(y,x) # if only it were true
likes(alaric, Anna Paquin) 

hence 

likes (Anna Paquin, alaric)


or something.

This kind of semantic parsing and extrapolation sounds alot like what
Tigre/Nomis (::scrs and bots passim) and Cyc [0] are trying to do.
However they've spent years and millions of $money trying to put in the
data. The code is quite simple so that's easy so what needs to be done
is seed it. And this would be done with people's 'web' (for want of a
better word) pages. Except then you have trust, scaling and DOS
problems. 

Does that sum it all up?

Do you see this Semantic web sitting along side the traditional Web? I
like my information to be human readable - I like sitting a reading a
news paper or, indeed, an email and the SemWeb seems counter productive
to this. Unless you plan to have parsing tools extract information from
human readable content.

Sorry, not really going anywhere with this, more just thinking aloud. 

Simon


[0] And the AI in the book I've just finished called Wisdom of
Crocodiles. Which is fantastic. And should be read by everybody. N.b
it's not about AIs and it's not set in the future. An AI like thing is
part of the plot though.