- From: Jan Algermissen <jalgermissen@topicmapping.com>
- Date: Wed, 4 Jan 2006 12:41:42 +0100
- To: <tim.glover@bt.com>
- Cc: <fmanola@acm.org>, <semantic-web@w3.org>
On Jan 4, 2006, at 12:15 PM, <tim.glover@bt.com> wrote:
> that in an ideal world data
> should be stored as triples
Yes, that is my thinking, too. Set aside issues of maturity of
technology, legacy software installations, etc...I do not see any[1]
benefit in using relational storage at all.
Sure, in the traditional relational world you need the relations to
bind keys and normal attributes together, but given global unique-
namespace keys (URIs) that binding becomes obsolete.
I say: "Down with the tyranny of class"[2] :o)
Jan
[1] *MAYBE* performance gains due to having attribute values that
usually come in groups on the same IO page, but there should be
possibilities to have the DBMS doing this grouping automatically
based on statistics (with OWL class definitions being used as hints
maybe)
[2] Taken from
"Emancipating Instances from the Tyranny of Classes in
Information Modeling"
Parsons, Wand; 2000 ACM TODS
________________________________________________________________________
_______________
Jan Algermissen, Consultant & Programmer
http://jalgermissen.com
Tugboat Consulting, 'Applying Web technology to enterprise IT'
http://www.tugboat.de
Received on Wednesday, 4 January 2006 11:41:48 UTC