W3C home > Mailing lists > Public > public-rdf-wg@w3.org > June 2011

Re: Need for a canonical byte stream for an RDF graph

From: Andy Seaborne <andy.seaborne@epimorphics.com>
Date: Tue, 28 Jun 2011 12:30:17 +0100
Message-ID: <4E09BB49.3080205@epimorphics.com>
To: public-rdf-wg@w3.org


On 28/06/11 09:21, Richard Cyganiak wrote:
> On 24 Jun 2011, at 00:52, Gavin Carothers wrote:
>>> Wouldn't the skolemization proposals make this a lot easier? Skolemize
>>> the blank nodes according to a deterministic pattern, then sort, then
>>> sign.
>>
>> That was one of my hopes early on following the skolemization work.
>> However as the current skolemization processes demands that two
>> systems must skolemize blank nodes to different URIs
>
> It doesn't demand that.
> http://dvcs.w3.org/hg/rdf/raw-file/default/rdf-concepts/index.html#section-skolemization
>
> Best,
> Richard

Right - so wouldn't workif you proceed as Ian describes, and publish the 
skolemized graph?  Publishing the skolemized form means the mapping from 
URIs to (local) bNodes is available.

This is a bit more general than signing the serialised form of the 
document as the signing is fixed by the serializiation order (after 
skolemization) for signing, not the document, but I can't think of a use 
case.

A more general form is to skolemize only the troublesome blank nodes, 
leaving e.g. leave the bNodes used in the object position only once 
(structured values).  These are not the "hard to label" ones because the 
there is a distinctive triple.  In fact, for data publishing the "hard 
to label" ones might be a sign that the data isn't web-friendly.

	Andy

>
>
>> this doesn't work
>> that well in practice. It may be that the skolemization process can be
>> defined in terms of c14n algorithm.
>>
>> Cheers,
>> Gavin
>>
>
>
Received on Tuesday, 28 June 2011 11:30:53 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:25:44 GMT