Re: XML Schema vs DAML/RDF/RDFS

On Wed, 17 Apr 2002, Dan Brickley wrote:

>
> On Wed, 17 Apr 2002, R.V.Guha wrote:
>
> Hi Guha,
>
> Interesting questions. Excuse the somewhat rambling response and
> telling you stuff you've yourself said at various points in the past...

[...]

> RDF is a pretty handy intermediate representation for data exchange. But
> I get the impression some folk feel it has been (to be blunt) side-tracked
> into the AI/KR world, and that implementors are now expected to implement
> everything in a logic programming / KR environment.

That feeling is very strong in many communities dealing with metadata.

> I think that's a mistaken view, and there are plenty of other
> deployment strategies, but we've not been that clear on the various
> options and tradeoffs available for implementors.

As you point out in the parts I've removed below, one isn't forced to
choose between a 'syntactic' or 'semantic' approach to a problem. Data may
be subject to various KR methodologies, and still in the end be accessed
through more conventional information retrieval methods. However, the
image one gets is that one is either a member of a semantic or a syntactic
community, and that one have to choose. Obviously that is false.

My own experiments with using RDF in a digital library context have
not been very successful. The problem isn't storing and exchange, but
querying. The reason for that seems (to me) to be that for any
data-structuring of any complexity, you will get a quite complicated
graph.

If you add heterogeneity to the data, you will have many types of graphs,
and you will need some sofisticated inferencing and recursive querying to
cope with it. I have not yet seen or heard of any tool that does this sort
querying automatically.

Indeed, the required effort for analysing the structure of all the graphs
seems (to me) larger or at its best equivalent with the syntactic effort
required to normalize incoming data using XSLT or XML a parser prior to
digesting stuff into a database. Having done the normalization, I can
choose between a range of fully fledged retrieval tools (rdbms or fulltext
engines depending on the data)...

I'm sorry, but at least currently, the semantic web has a very weak case
for use in a information retrieval context.


Sigge

Received on Thursday, 18 April 2002 06:55:19 UTC