W3C home > Mailing lists > Public > semantic-web@w3.org > August 2006

Re: RDF vs RDBMS

From: Sampo Syreeni <decoy@iki.fi>
Date: Tue, 29 Aug 2006 19:44:28 +0300 (EEST)
To: Danny Ayers <danny.ayers@gmail.com>
cc: semantic_web@googlegroups.com, Semantic web list <semantic-web@w3.org>
Message-ID: <Pine.SOL.4.62.0608291903500.13314@kruuna.helsinki.fi>

On 2006-08-29, Danny Ayers wrote:

> Where things start to get really interesting is that the predicates
> can appear in "tables" too: [...]
>
> At this point it may be easier to stop thinking in terms of the 
> relational model, the object-oriented model - the inheritance bits at 
> least - is probably closer conceptually (though still very different).

Actually even this isn't too far removed from relational practice: the 
catalog of an RDBMS contains precisely this sort of meta-information. 
Really the only difference is that in the RDF world, this 
schema/semantics/metamodel stuff was standardized early on as RDFS and 
OWL, whereas in the relational world standard catalog formats have been 
a long time coming and still aren't well integrated with the bare bones 
tabular machinery (e.g. the ISO SQL92 standard catalog is read-only).

There's probably no single, best way to relate RDF's 
graph/semistructured data model to the standard relational formalism. My 
own understanding of the concepts is based on viewing the set of triples 
as a universal relation where column labels have been unpivoted downto a 
single property column, the subject takes the place of a surrogate key, 
the field values are stored as the object, and blank nodes correspond to 
marked nulls. After that most of the storage, mapping and 
interoperability issues can be reinterpreted in terms of the 
physical/logical dichotomy: your mapping into multiple binary relations 
would simply be an instantiation of the universal relation as a column 
store, the many object-relational mappings out there can be used to give 
multiple different realizations of a single inheritance hierarchy as 
long as the enforced data dependencies stay the same, and so on.

The point is that the difference in data models is probably more about 
viewpoint than fundamental power of representation or anything like 
that.
-- 
Sampo Syreeni, aka decoy - mailto:decoy@iki.fi, tel:+358-50-5756111
student/math+cs/helsinki university, http://www.iki.fi/~decoy/front
openpgp: 050985C2/025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2
Received on Tuesday, 29 August 2006 16:45:08 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:47:17 UTC