Re: Data stuff

Deborah L. McGuinness wrote:
> 
> Bijan Parsia wrote:
> 
>>
>> The owl working group will be discussing datatypes and n-ary data 
>> predicates today at the f2f.
>>
>> See:
>>    http://www.w3.org/2007/OWL/meeting/2008-07-16#Normative_datatypes
>>    http://www.w3.org/2007/OWL/wiki/N-ary_Data_predicate_use_case
>>    http://www.w3.org/2007/OWL/wiki/N-ary_Data_predicate_proposal
>>
>> If you have use cases, strong feelings, questions, etc. now is a good 
>> time to voice them.
>>
>> Cheers,
>> Bijan.
>>
>>
> thx for the reminder.
> two points arise for me:
> 1 - i do not see datetime on the agenda - is it embedded in something i 
> missed or are we holding off because it needs more discussion

It is there http://www.w3.org/2007/OWL/wiki/F2F3_Agenda as
"Date and time types" just before N-ary within the Datatypes topic.
We are now on break.

> 2 - if datetime is discussed, i would have trouble living with a 
> solution that forced me to represent datetime as structured owl objects 
> rather than xsd:datetime or something like a more granular xsd:datetime 

My write-up of the current proposal uses xsd:dateTime.  There will be
some hacks required for dealing with things that differ from full 
xsd:dateTime with a timezone component.

> because of the sheer volume of datetime data my applications have.
> exemplar use case for datetime is:
> 2a. retrieve data with a start time of yymmdd for a duration of 10 days  
> (optional time zone, default is ut, optional hours, minutes, seconds)

I strongly dislike quietly defaulting to UTC.  Tools should at least
complain when they do this.  IMHO it would be better to just disallow
this and expect tools to either gag on it or interact with the user to
repair it.

> 3 - i notice that comparison is embedded in the proposal.  as long as i 
> get something that lets me do comparisons and numerical ranges i can 
> make due. exemplar use cases are:
> 3a. atmosphere above xxx feet and below yyy feet
> 3b. high geomagnetic activity = something with a kp index above 7
> 
> 

Currently spec'd datatype restrictions support this don't they?

-Evan

Received on Monday, 28 July 2008 15:00:14 UTC