[Fwd: RE: Issue for Device Independence Working Group: URIs for units and MIME types]

Enclosed is a message that was cc-ed to me as SWBPD chair. It concerns 
the need for a "units" ontology in the context of the work of the W3C 
Device Independence Working Group.

I welcome discussion on this topic to establish whther we can be of any 
help.

Guus

-----Original Message-----
From: Williams, Stuart [mailto:skw@hp.com]
Sent: 30 June 2004 13:05
To: Rhys Lewis
Cc: ian@w3.org; W3C DIWG (E-mail); tag@w3.org; Guus Schreiber
Subject: RE: Issue for Device Independence Working Group: URIs for units
and MIME types


Hello Rhys,

Summary:

1) URI's for media-types is the subject of TAG issue uriMediaType-9.
2) The question of URI naming practice for physical units might be better
directed at the Semantic Web Best Practices WG (cc'd WG chair)

More comments in line, below...

> -----Original Message-----
> From: Rhys Lewis [mailto:rhys.lewis@volantis.com] 
> Sent: 30 June 2004 08:48
> To: skw@hp.com
> Cc: ian@w3.org; W3C DIWG (E-mail)
> Subject: Issue for Device Independence Working Group: URIs 
> for units and MIME types
> 
> Hello Stuart,
> 
> Ian Jacobs suggested that we shoud write to you about a 
> particular issue that we uncovered at the Device Independence 
> Working Group face to face meeting last week. It concerns the 
> representation of the values of properties such as MIME type 
> and unit of measurement.
> 
> This question arises in the context of the work we are doing 
> to define a vocabulary for the characteristics of devices 
> used to access the Web. DIWG has various work items relating 
> to this vocabulary and its representation, for example within CC/PP.
> 
> We are trying to define an RDF vocabulary for these 
> characteristics. We decided that it would be better to use 
> URIs rather than string literals for property values that 
> have special meaning wherever possible. For example, we'd 
> prefer to use URIs to identify the units (inch, centimetre, 
> foot) associated with a device characteristic. We'd also like 
> to be able to use URIs to identify MIME types for consistency 
> in our vocabulary.

With respect to measurement units (time, length, mass...) it seems to me
that you are/or should be looking for an existing ontology of URI for naming
such things. Off the top of my head I don't know of such a thing, but I'd
suggest consulting the Semantic Web Best Practices WG who may be able to
help with some of this. There are a few at
<http://sweet.jpl.nasa.gov/ontology> that might be of interest... but I
think that the SWBP-WG would give you the best advise - and I'd expect it to
promote reuse over re-invention.

> We have a couple of initial questions about this issue and we 
> would really appreciate help from the TAG.
> 
> First, is this use of URIs sensible, or should we simply use 
> string representations for things as 'fundamental' as MIME 
> types and units?

I'm pretty confident that the TAG would agree that this is a sensible use of
URI (tho' I have not asked them specifically). Our disposition generally is
in favour of the use of URI for naming things, and preferably unamiguously
naming things - real-world things, abstract things and computational objects
- basically any kind of thing.

The TAG does have an open issue on the use of URI to name MIME types [1].
Registered Internet Media Types are catalogued in the IANA registry at
<http://www.iana.org/assignments/media-types/> such that the media-type
registration for the media-type of say, image/png can be found at
<http://www.iana.org/assignments/media-types/image/png> (this mapping is not
univerally deployed - some obvious deference attempts - eg for text/plain -
end in 404's). I believe that we (the TAG) would like to see URI of this
from being used to name internet media types. An issue that we have been
discussing with the IETF is that IANA/IETF are unable to guarantee the
persitence of these URI. Reorganisation of the IANA site might yield a
different 'prefix' ahead of the media-type. Roy Fielding has suggested
thinking of media-type names as relative URI - the problem being a lack of
commitment to the persistence of the associated base URI. You can find our
most recent exchange with IANA/IETF at [2]... although it is somewhat old
and AFAIK lacking in a response.

[1] http://www.w3.org/2001/tag/issues#uriMediaType-9
[2] http://lists.w3.org/Archives/Public/www-tag/2003Feb/0302.html

> For us, it seems URIs are the right approach 
> for generality and consistency. In addition string 
> representations have some small problems of their own. For 
> example, I spell the SI unit of length 'metre', but then I 
> was educated in England. Not everyone has the same view.

I think the TAG would agree (on the use of URI names).... and again one of
the SWEET ontologies might interest you (units.owl), but I'd also check
best-practice with the SWBP-WG.

> Second, if URIs are the right way to represent these values, 
> should there be a standard set for things like MIME types 
> that are of general utility for W3C? 

Definitely a BPWG question. I think there is an issue of reuse v
reinvention. Personnally I think it would be problematic to try to 'clone'
or 'duplicate' a number of general ontologies in W3C space. Consistent reuse
of generally accepted 'good' ones would reinforce the 'network effect' and
build the sense of concensus around a particular ontology.

> We note that other 
> groups are currently defining properties that can contain 
> MIME types. In particular, the working draft "Assigning Media 
> Types to Binary Data in XML" 
> 
> (http://www.w3.org/TR/2004/WD-xml-media-types-20040608/#contentType) 
> 
> specifies use of the name of the IANA media type token rather 
> than a URI for this purpose.

Hmmm.... whilst ideally there would be a stable URI to leverage, I think
that we have not managed to establish that IANA will actually take that on
which is a pity. I like Roy's idea of thinking of media-types as relative
URI references - at least that would admit them as "AnyURI" values - and
leaves room for an absolute URI as and when a persistent base URI can be
agreed.

> Would it be possible for us to obtain some assistance from 
> the TAG on these questions?

I think the uriMediaType-9 issue covers one of your questions and on the
other I think your best advise would come from SWBP-WG.

I hope this is of some help. It may be that there is a focussed question
here that the TAG could address. But on the generic question of "use URI's"
I believe that the TAG would agree with you. The harder question is... "ok
then... which ones?"

> Best wishes
> 
> Rhys Lewis

Best regards

Stuart Williams




-- 
Free University Amsterdam, Computer Science
De Boelelaan 1081a, 1081 HV Amsterdam, The Netherlands
Tel: +31 20 444 7739/7718
E-mail: schreiber@cs.vu.nl
Home page: http://www.cs.vu.nl/~guus/

Received on Monday, 5 July 2004 08:22:15 UTC