W3C home > Mailing lists > Public > www-rdf-interest@w3.org > November 1999

Re: Storing RDF in relational databases

From: Sergey Melnik <melnik@DB.Stanford.EDU>
Date: Sun, 14 Nov 1999 19:00:10 -0800
Message-ID: <382F773A.153E784F@db.stanford.edu>
To: Jonas Liljegren <jonas@paranormal.o.se>
CC: RDF Interest Group <www-rdf-interest@w3.org>
Hi Jonas,

thanks for your detailed information! I updated the summary page.

Could you cut-and-paste a couple of SQL statements that you use to query
the DB? For example, how do you select all triples with a given subject
(think of reification)? If it is not possible in one statement, please
write down all of them. I'm trying to figure out what the typical usage
patterns of a DB-based solution are and how expensive it is in terms of
the number and complexity of queries.

Could you give an example of IDs that you generate for the reified
statements?

As to splitting URIs into namespaces and names: first, a clear splitting
is not possible given just the URI. However, it doesn't really matter
since it is an internal implementation issues. After importing 6 mio
triples from the Open Directory (OD) into the database using the schema
[1] I found out that the table "namespaces" was about 50% of the size of
"resources". The problem is that many URIs in OD are arbitrary URLs, so
the savings are minimal. And my prediction is that this ratio will be
similar even if I get more RDF schema-like data into the DB, just due to
the fact that there are so many different entities around. I'm planning
to get rid of the table "namespaces". You probably wouldn't save
anything by having it.

So far,
Sergey
Received on Sunday, 14 November 1999 21:55:38 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:42 GMT