Databases on the Semantic Web

hi All,

I have a question about the exposure of databases to the Semantic Web.
Basically, I would like to know your opinion: What do you think, will
semantic annotations (encoded in RDF) be used to

(a) expose only the (semantics of a) database schema (and provide a
description how to access it)

or will it be used to

(b) expose (the semantics of) both database schemata and *database data* ?


Which variant is more reasonable?

Here my opinion:

Variant (a) will need some description format like DAML-S to describe the
means to access the databases. Variant (a) has the advantage that access
control etc. may be handled more elegantly than in (b). However, reasoning
over such a database will by very hard, because you first have to fire the
"right" query to get the "right" data (which then might be asserted to the
knowledge base of the agent). At least to me, it seems that inference
between several databases can by very hard (impossible??) this way.

Variant (b) has the advantage that an agent(logical reasoner) could
_directly_ access and reason about _all_ data, thus enabling inference
between _multiple_ databases. However, it has the disadvantage that one has
to dump that whole database to RDF periodically.

What do you think?
Can't wait to hear your findings and opinions!


Joachim


PS: I have read the article on exposure of database to the Semantic Web[1]
and I experimented with the implementation in [2] - does anyone have
additional ressources on that topic?


[1] Tim Berners-Lee. "Relational Databases on the Semantic Web"
http://www.w3.org/DesignIssues/RDB-RDF.html

[2] DbView.py - "view an SQL DB thru RDF glasses", Open Source
implementation by Dan Connolly
http://www.w3.org/2000/10/swap/dbork/dbview.py


Joachim Peer
Research Assistant
Institute for Media and Communications Management
Blumenbergplatz 9
CH-9000 St. Gallen
Tel.: ++41 (0)71 224 3441

Received on Tuesday, 12 March 2002 00:03:55 UTC