Re: RDB2RDF Working Group

Marcelo,

In my view you should definitely look at OWL 2 and not only OWL in this
respect. First  of all, there are some new type of restrictions that may
be relevant (qualified cardinality restrictions, for example; ie, in
Birte's example, you can replace Thing with something more specific,
something OWL 1 did not allow), there are better tools to handle
datatypes (eg, datatype restrictions, a really powerful stuff), and,
maybe the most relevant and important for you guys, the definition of
keys. One can say things like a person is uniquely identified by his/her
name _and_ email address, for example.

But all this is a bit digression on SPARQL. The only place it comes in
is that in SPARQL 1.1, if everything goes as planned, endpoints may
advertise that they can handle OWL entailments, ie, they can consider
such restrictions.

Cheers

ivan

Marcelo Arenas wrote:
> Hi Birte,
> 
> Thanks for your note, it is definitely relevant for what we are doing.
> Actually, we are currently discussing some alternative approaches to
> map the data definition language of SQL into RDF and OWL:
> 
> http://www.w3.org/2001/sw/rdb2rdf/wiki/Initial_Round_of_Presentations
> 
> In particular, Juan Sequeda gave a talk today in our weekly meeting
> where he presented an approach for translating relational databases
> that may include keys and foreign keys.
> 
> Cheers,
> 
> Marcelo
> 
> On Mon, Nov 23, 2009 at 3:01 PM, Birte Glimm
> <birte.glimm@comlab.ox.ac.uk> wrote:
>> Marcelo,
>> I am not sure that the following is of any relevance, but I thought I
>> mention it anyway ;-)
>> To some extend OWL has the power to express key constraints etc. For
>> example, you can say that each Person must have exactly one social
>> security number:
>> SubClassOf(Person ObjectExcactCardinality(1 hasSSN owl:Thing))
>> or in clause form:
>> Person(x) and hasSSN(x, y1) and hasSSN(x, y2) -> y1=y2
>> Person(x) -> hasSSN(x, f(x)) for f(x) some skolem constant.
>> The limiting factor in OWL versus RDBs is more that OWL can (natively
>> without reification) only express binary relations whereas DBs work
>> over n-ary relations. This might render OWL's abilities not suitable
>> for what you want.
>> Anyway, I think it is an interesting topic and I hope I can keep a bit
>> up todate about what the RDB2RDF group is doing.
>> Birte
>>
>>
>> 2009/11/23 Axel Polleres <axel.polleres@deri.org>:
>>> FYI, some conversation with Marcelo Arenas from the RDB2RDF working group which I share with his permission...
>>>
>>> Axel
>>>
>>> Begin forwarded message:
>>>
>>>> From: Axel Polleres <axel.polleres@deri.org>
>>>> Date: 23 November 2009 15:40:50 GMT+01:00
>>>> To: "Marcelo Arenas" <marcelo.arenas1@gmail.com>
>>>> Subject: Re: RDB2RDF Working Group
>>>>
>>>> Hi Marcelo!
>>>>
>>>> Good to hear from you!
>>>>
>>>> First of all, can I share this conversation with the SPARQL WG/Chairs?
>>>> (I ask because you addressed to me alone)
>>>>
>>>> Now my answer:
>>>>
>>>> Looking forward to collaborate with you on the RDB2RDF side of the liaison with SPARQL...
>>>> Indeed the definition of integrity constraints such as keys and foreign keys is not a part of SPARQL.
>>>> In fact, SPARQL is not a data definition language: we define a query language (SPARQL/query [1]) and
>>>> a data manipulation language (SPARQL/update [2]), but aren't chartered for designing data definitions...
>>>>
>>>> Dataset descriptions, as far as we are concerned with (as a part of SPARQL/service descriptions [3])
>>>> will be very basic, just providing hooks to data descriptions at most, that is a minimal set of properties to link to dataset descriptions in some external formalism (concretely void and saddle have been discussed earlier,
>>>> but we will most likely not actively promote a particular data definition language).
>>>>
>>>> I hope that clarifies matters. in fact, I'd honestly see the definition/description of constraints on datasets as something on the side of extensions of OWL/RDF Schema rather than on the side of SPARQL. That latter sentence is my personal view... probably something to be discussed/suggested in the upcoming "next steps on RDF" W3C workshop.
>>>>
>>>> best regards,
>>>> Axel
>>>>
>>>> 1. http://www.w3.org/TR/sparql11-query/
>>>> 2. http://www.w3.org/TR/sparql11-update/
>>>> 3. http://www.w3.org/TR/sparql11-service-description/
>>>>
>>>>
>>>>
>>>> On 23 Nov 2009, at 15:29, Marcelo Arenas wrote:
>>>>
>>>>> Dear Axel,
>>>>>
>>>>> As you probably know, there is a new working group in the W3C whose
>>>>> mission is to standardize a language for mapping relational data into
>>>>> RDF and OWL  (RDB2RDF WG). I am an invited expert in that group.
>>>>>
>>>>> The work on SPARQL is fundamental for the work of the RDB2RDF WG, so
>>>>> we would like to coordinate our efforts with that of the SPARQL WG.
>>>>> For that reason, I have volunteered to be the liaison between the
>>>>> RDB2RDF group and your group.
>>>>>
>>>>> Currently, we are trying to define the list of features of the data
>>>>> definition language of SQL that will be supported in the mapping
>>>>> language for relational data into RDF and OWL. One of these features
>>>>> is the definition of integrity constraints such as keys and foreign
>>>>> keys, so I was wondering whether you have discussed about the
>>>>> possibility of including integrity constraints in the data definition
>>>>> language of SPARQL. I took a look at the SPARQL WG Wiki and I couldn't
>>>>> find any reference about this, but maybe I am missing something. Thank
>>>>> you in advance for any information about this.
>>>>>
>>>>> Cheers,
>>>>>
>>>>> Marcelo
>>>>>
>>>
>>>
>>
>>
>> --
>> Dr. Birte Glimm, Room 306
>> Computing Laboratory
>> Parks Road
>> Oxford
>> OX1 3QD
>> United Kingdom
>> +44 (0)1865 283529
>>
> 

-- 

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