W3C home > Mailing lists > Public > www-rdf-interest@w3.org > March 2004

Re: a bnode URI scheme?!

From: Andrew Newman <andrew@tucanatech.com>
Date: Mon, 15 Mar 2004 17:28:29 +1000
Message-ID: <40555B1D.1070308@tucanatech.com>
To: www-rdf-interest@w3.org

Geoff Chappell wrote:
> RDF Gateway has a merge method on its datasource object that allows you
> to merge one set of triples with another, removing redundant triples in
> the process. For method docs w/example see:
> 
> http://www.intellidimension.com/pages/rdfgateway/reference/script/dataso
> urce_merge.rsp
> 
> It's intended to deal with the problem you describe below, more than it
> is to reduce a graph to lean form.
>  

Isn't that nearly always going to be wrong?  I mean just because an 
bnode has the same properties (say first name and last name or even just 
first name) doesn't mean they are the same thing.

> 
>>A use-case where this is needed is the versioning of rdf models -- as
> 
> a
> 
>>model changes over time and a version control system merges the
> 
> changes
> 
>>in, bNodes will proliferate (in particular list bNodes created by
>>parseType="collection" ). I'm actually encountering this issue right
> 
> now
> 
>>with Rhizome, my content management system that treats everything as
> 
> RDF
> 
>>(rx4rdf.sf.net).
>>

There are other ways to version RDF models.  There's basically three 
levels of grouping that I'm aware of:

1/ Dropping and adding whole models.
2/ Constructing larger models from unions and intersections of others.
3/ Selecting a single resource, by its properties, then removing it.

These should probably all be done in transactions too.

In other use cases I think it would be *more* important to keep as many 
statements as possible not throw them away.
Received on Monday, 15 March 2004 02:29:32 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 18 February 2014 13:20:07 UTC