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

Dataset semantics

From: Antoine Zimmermann <antoine.zimmermann@emse.fr>
Date: Tue, 18 Oct 2011 17:25:38 +0200
Message-ID: <4E9D9A72.7030409@emse.fr>
To: public-rdf-wg@w3.org
Dear all,

I was overloaded with other things and haven't participated in the 
recent discussions around graph and dataset semantics but I am now 
catching up and these are some thoughts I have after reading all that 
was said. I'd like to get your opinion on this.

These are some things I expect regarding the semantics of datasets:
  1. They should not impact the semantics of RDF;
  2. There is no single fit-for-all semantics of datasets so the 
semantics must be flexible to accomodate use cases and existing 
  3. There should exist a way to tell the world what specific semantics 
one application is using.

Item 1 implies that the semantics of datasets are distinct from the 
semantics of RDF.
Item 2 implies that different conformant applications may entail 
different things. Yet, there should be common requirements to make a 
semantics "conformant".
Item 3 is rather open but I am thinking of something like a description 
of a dataset that says, e.g., "bnodes are locally scoped" or (following 
Sandro's proposal) "graph labels are used to RDF-denote the graph 
containers" or "the default graph is the union of the named graphs" or 
"the default graph represents shared knowledge, valid in all named 
graphs", etc... All these statements say something about the semantics 
of datasets that is used in a system.

Regarding the minimal requirements for a semantics of datasets, there is 
already a proposal [1]. I updated it to make it clearer. That proposal 
addresses many of the issues that Richard puts forward in a recent email 
[2]. In particular, it says that what is "true in a graph" (or, in a 
context) is not necessarily true in a different graph (or context). This 
minimal semantics does not provide any means of talking about an RDF 
Graph. This should be addressed by the mechanism that I envision in Item 
3 above.

[1] Semantics (in TF-Graphs/RDF-Datasets-Proposal). 
[2] Datasets and contextual/temporal semantics. 
Antoine Zimmermann
ISCOD / LSTI - Institut Henri Fayol
École Nationale Supérieure des Mines de Saint-Étienne
158 cours Fauriel
42023 Saint-Étienne Cedex 2
Tél:+33(0)4 77 42 66 03
Fax:+33(0)4 77 42 66 66
Received on Tuesday, 18 October 2011 15:26:08 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:04:09 UTC