Re: PROV-ISSUE-493: prov:type has type Value; valid values too general, include number, datetime, boolean, etc. [prov-dm]

On Sep 12, 2012, at 7:22 AM, Luc Moreau <l.moreau@ecs.soton.ac.uk> wrote:

> Hi Curt and Stephan,
> 
> I am less certain about this change.
> 
> First, do you mean QName as in xsd:QName?

Yes.

> Why not use the prov:QualifiedName, which we already have (and can be transformed into uris).

We could do that, what is the distinction?  I suggested xsd:QName since we were already using it for the id attribute in the XML schema.

> 
> But then, why just prov:QualifiedName , and why not URI (xsd:anyURI)?

We use URI's for types in PROV-O so I am not against this.  I suggested xsd:QNames since we were already using it with ids, and because it is a currently valid prov:type type since QName is mentioned as a valid value of a DM Value.

> 
> The reason why this was left unspecified is that PROV, intentionally, refrained from defining
> what a type system is, and therefore, a consequence, was that we didn't define how to
> represent a given type value.

Then perhaps the question should be, "How do we define a type system for PROV-XML that makes sense?".  I think the type system PROV-O adopted is very good, and the type system the DM defines is too broad.

--Stephan

> 
> Luc
> 
> On 09/12/2012 01:27 PM, Curt Tilmes wrote:
>> 
>> I agree with Stephan.  The real reason for having prov:type at all is
>> to encourage consistency.  Qnames encourage capturing semantic meaning
>> beyond free text.
>> 
>> The types we've defined
>>  http://www.w3.org/TR/prov-dm/#term-attribute-type
>> set a precedent for the type of types we think should fill prov:type,
>> and the discussion of prov:type in the extensibility points section:
>>  http://www.w3.org/TR/prov-dm/#extensibility-section
>> shows examples defining new prov:types as qnames in other namespaces.
>> 
>> This would require some rework of examples, but I think the change
>> would be valuable in the long term.
>> 
>> Curt
>> 
>> On 09/12/2012 02:19 AM, Stephan Zednik wrote:
>>> A quick reminder about this issue.
>>> 
>>> Looking at the PROV-DM document again I see a few examples where simple
>>> non-qname strings are used for prov:type values.
>>> 
>>> From example 21 (http://www.w3.org/TR/prov-dm/#anexample-communication)
>>> 
>>> prov:type="fine paying, check writing, and mailing"
>>> 
>>> I think in most if not all of these cases the prov:type value could be
>>> simplified to a qname.
>>> 
>>> I understand this change is significant due to the timing of the
>>> suggestion, but I believe the benefit of making this change is worthwhile.
>>> 
>>> Thanks,
>>> --Stephan
>>> 
>>> On Sep 4, 2012, at 11:18 AM, Provenance Working Group Issue Tracker
>>> <sysbot+tracker@w3.org <mailto:sysbot+tracker@w3.org>> wrote:
>>> 
>>>> PROV-ISSUE-493: prov:type has type Value; valid values too general,
>>>> include number, datetime, boolean, etc. [prov-dm]
>>>> 
>>>> http://www.w3.org/2011/prov/track/issues/493
>>>> 
>>>> Raised by: Stephan Zednik
>>>> On product: prov-dm
>>>> 
>>>> The value of prov:type is a Value
>>>> (http://www.w3.org/TR/prov-dm/#term-value) which has the following
>>>> definition:
>>>> 
>>>> A value ◊ is a constant such as a string, number, time, qualified
>>>> name, IRI, and encoded binary data, whose interpretation is outside
>>>> the scope of PROV. Values can occur in attribute-value pairs.
>>>> 
>>>> Each kind of such values is called a datatype. Use of the following
>>>> data types is recommended.
>>>> 
>>>> The RDF-compatible [RDF-CONCEPTS] types, including those taken from
>>>> the set of XML Schema Datatypes [XMLSCHEMA11-2];
>>>> Qualified names introduced in this specification.
>>>> The normative definitions of these datatypes are provided by their
>>>> respective specifications.
>>>> 
>>>> This means that numbers, datetimes, booleans, and unstructured strings
>>>> are valid values of prov:type.  The prov value section on RDF
>>>> compliance also seems to suggest there should be a prov:type datatype
>>>> property in prov-o, which to my knowledge does not currently exist.
>>>> 
>>>> So my question is, are we ok with numbers, datetimes, booleans as
>>>> valid values of prov:type?  All of the examples in the DM document
>>>> appear to use qnames for values of prov:type.
>>>> 
>>>> Second, is there support for a proposal to restrict values of
>>>> prov:type to qnames?
>>>> 
>>>> 
>>>> 
>>>> 
>>> 
>> 
>> 
> 
> -- 
> Professor Luc Moreau
> Electronics and Computer Science   tel:   +44 23 8059 4487
> University of Southampton          fax:   +44 23 8059 2865
> Southampton SO17 1BJ               email: l.moreau@ecs.soton.ac.uk
> United Kingdom                     http://www.ecs.soton.ac.uk/~lavm
> 
> 
> 

Received on Wednesday, 12 September 2012 17:30:49 UTC