RE: BSWL - Basic Semantic Web Language

Sean B. Palmer wrote
>
>
> I'm pleased to announce a new proposal for a stripped down RDF-in-XML
> syntax called "BSWL" or the "Basic Semantic Web Language". The
> proposal is available on the Web at:-
>
>    http://purl.org/net/bswl
>
> Briefly, the features advantages over RDF M&S are:-
>
> * Simpler syntax - no typed or anonymous nodes, allows one to form
> triples simply by nesting QNames

I generally like it.

> * Allows you to nest triples so that the subject of the former triple
> becomes the object of the latter triple

This, particularly, is very nice.

> * Is possible to convert back into RDF M&S, and vice versa (once RDF
> Core decide what to do about anonymous nodes)
> * Files using abbreviated BSWL tend to be shorter than RDF M&S

Might you convert some number of the M&S examples to your language so we
might compare?

>
> An example of a BSWL chunk is:-
>
> <bswl:t qname="Sean">
>    <bswl:po qname="likes">
>      <bswl:o qname="TheSimpsons"/>
>    </bswl:po>
> </bswl:t>
>
> Which can be shortened using abbreviated BSWL into:-
>
> <Sean>
>    <likes>
>      <TheSimpsons/>
>    </likes>
> </Sean>
>
> Further details are available in the proposal itself.

The inability to deal with 'anon' nodes is a problem however. Not that we
need to keep the concept of anon nodes _as is_ rather the use of anon nodes
to represent quantified variables e.g.:

:Sean :likes [ :called "Chocolate" ] .

really means:

exists ?x such that [?x :called "Chocolate"] and [:Sean :likes ?x]

in your pseudo syntax

<Sean>
	<likes>
		<x:variable name="?x">
			<called>
				<Chocolate>
			</called>
		</x:variable>
	</likes>
</Sean>

-Jonathan

Received on Tuesday, 17 July 2001 08:27:20 UTC