Re: Operations on RDF datasets (ISSUE-111)

On Jan 17, 2013, at 12:14 , Andy Seaborne <andy.seaborne@epimorphics.com> wrote:

> 
> 
> On 17/01/13 10:51, Ivan Herman wrote:
>> 
>> On Jan 16, 2013, at 21:00 , Andy Seaborne <andy.seaborne@epimorphics.com> wrote:
>> 
>>> 
>>> 
>>> On 16/01/13 18:10, Richard Cyganiak wrote:
>>>> As discussed in the call today, we need to decide whether to define any operations on RDF Datasets in the RDF Concepts document.
>>>> 
>>>> There is a wiki page that lists some possible candidate operations, along with some discussion, pointers to possible definitions, etc.:
>>>> http://www.w3.org/2011/rdf-wg/wiki/Dataset_Operations
>>>> 
>>>> The candidate operations listed there are:
>>>> 
>>>> * Isomorphism
>>> 
>>> Include for testing otherwise no need.
>> 
>> I am not sure I agree with 'no need': the RDF Semantics does define the equivalence of graphs, I think having a clear definition extended to dataset is, sort of, missing.
>> 
>> Anyway, whatever the reason, I believe this one is a +1 for me.
> 
> equivalence/isomorphism is the one case where we might get agreement and it is needed testing.  Let's consider this one separately to the others.

Agreed.

> 
>>> The defn quoted in the telecon is fine but clearer if it says "there is a single isomorphism mapping for the dataset", not a separate mapping per graph.
>>> 
>>> (which makes it an isomorphism from one set mixing quads and triples to another set)
>>> 
>>>> * Union
>>>> * Merge
>>>> * Untrusting Merge
>>>> * Equivalence
>>>> * Entailment
>>>> * Equality
>>>> * Folding/Unfolding
>>>> * Union Dataset and Merge Dataset
>>> 
>>> No need.
>>> 
>>> (reason - we don't have time nor want to reopen the dataset semantics discussions)
>>> 
>> 
>> I am indeed a bit concerned about reopening discussions here. There
>> is
> also the argument that we should not add things to the standard for
> which there is no stable community experience yet.
>> 
>> However (without using telcon time) maybe it is worth considering
> adding these into the note on Datasets that Antoine is writing. If there
> is a group who is interested in finding a proper definition for those,
> that is.
>> 
>> Ivan
> 
> Are you suggesting all of the operations (sans iso/equiv) in that note?

Not necessarily. My point was (but it was not clear, sorry about that) not to even try to define these things for the purpose of being added to a Rec (eg, using telco time, etc). All these should be considered only as a possible inclusion in the note (if there is an agreement on the definitions)

Ivan

> 
> The usage of labelling of URI for the named graph in each dataset may be different; it may be location snapshots at different times, etc etc. Groundhog Day.
> 
> (We might agree on definitions of syntactic operations and say "if the labelling style is the same but then we have to agree on whether it is the "right" operation to highlight.)
> 
> 	Andy
> 
>> 
>>>> 
>>>> Any opinions? In particular, arguments *for* including some of this stuff? Candidate definitions, etc.?
>>>> 
>>>> Best,
>>>> Richard
>>>> 
>>> 
>> 
>> 
>> ----
>> Ivan Herman, W3C Semantic Web Activity Lead
>> Home: http://www.w3.org/People/Ivan/
>> mobile: +31-641044153
>> FOAF: http://www.ivan-herman.net/foaf.rdf
>> 
>> 
>> 
>> 
>> 


----
Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
FOAF: http://www.ivan-herman.net/foaf.rdf