Re: Is there an ISO 8601 based alternative to accrualPeriodicity in DCAT?

Philip,

Thanks for the background! It’s great to see that POD is run using publicly visible infrastructure like GitHub.

It might also be worth asking the question on the Dublin Core Vocabulary mailing list [1]. Ultimately, DCAT just re-uses a lot of properties from Dublin Core, and Dublin Core is the ultimate authority on their semantics.

Best,
Richard

[1] https://www.jiscmail.ac.uk/cgi-bin/webadmin?A0=DC-VOCABULARY


On 10 Jun 2014, at 18:20, Philip Ashlock - XI <philip.ashlock@gsa.gov> wrote:

> Thanks. That makes sense. For what it's worth, I can imagine using 8601 with only the repeating interval syntax (eg R/P1Y2M10DT2H30M) but not the date (eg R/2008-03-01T13:00:00Z/P1Y2M10DT2H30M). Though as you said that's not intuitive and would likely imply that full ISO syntax would be acceptable. 
> 
> The context for this is the Project Open Data (POD) schema which has tried to align with DCAT. The current version of the POD schema was finalized before the current version of DCAT stabilized, but there is work on an updated version of the POD schema so there might be opportunities to better align them. You can see a crosswalk at:
> 
> http://project-open-data.github.io/metadata-resources/
> 
> The question about accrualPeriodicity came up on this thread (https://github.com/project-open-data/project-open-data.github.io/issues/292) and it sounds like there's interest to using 8601 since it's already used by other properties and provides this flexibility. Then the question is whether to continue using that with the accrualPeriodicity term or use another term. It sounds like your suggestion is to use a different term, likely a new one. 
> 
> 
> On Tue, Jun 10, 2014 at 12:55 PM, Richard Cyganiak <richard@cyganiak.de> wrote:
> Hi Philip,
> 
> On 10 Jun 2014, at 17:18, Philip Ashlock - XI <philip.ashlock@gsa.gov> wrote:
> > So are you saying an 8601 formatted recurring date would fit within the range of dctype:Frequency but only if it didn't include a start or end date?
> 
> I’m concluding this simply from the description of dctype:Frequency: “A rate at which something recurs.” Something like “Monthly” is a “rate”. Something like “On the 27th of July, August and September 2014” isn’t, strictly speaking.
> 
> > The requirement is both to express other frequencies than cld:Frequency as well as more precise times.
> 
> I see.
> 
> Let me put it this way.
> 
> Nobody can stop you from using any value you’d like with dcterms:accrualPeriodicity. The interesting question though is if you will see interoperability.
> 
> There’s a good chance that another DCAT user who understands accrualPeriodicity would understand the cld:Frequency values, because they are sort of the obvious values to use with these properties.
> 
> The chance that they would understand the ISO 8601 values would be lower. If those ISO 8601 values include not just a frequency but also a start and/or end date, then they might even be confused, because that’s not what you’d expect from reading the description of accrualPeriodicity and dcterms:Frequency.
> 
> Therefore, in that case, it’s probably best to use a different property. You might have to define that property yourself if nothing suitable exists out there.
> 
> Best,
> Richard
> 
> 
> 
> >
> >
> > On Tue, Jun 10, 2014 at 7:29 AM, Richard Cyganiak <richard@cyganiak.de> wrote:
> > Hi Philip,
> >
> > On 9 Jun 2014, at 23:22, Philip Ashlock - XI <philip.ashlock@gsa.gov> wrote:
> >
> > > As far as I can tell accrualPeriodicity is limited to the predefined terms for frequency http://dublincore.org/groups/collections/frequency/
> >
> > I don’t think that’s the case. The range of dcterms:accrualPeriodicity is dctype:Frequency. It doesn’t follow that values are limited to cld:Frequency.
> >
> > (However, I do agree that a recurring interval with a start date is out of range for dctype:Frequency.)
> >
> > > Are there any existing terms that could be used in place of accrualPeriodicity using 8601 recurring interval syntax for more flexibility?
> >
> > I’m not aware of anything suitable.
> >
> > What is the requirement here? Express other frequencies besides those in cld:Frequency? Or express the precise time (e.g., not just monthly, but monthly on the 10th)?
> >
> > Best,
> > Richard
> >
> >
> >
> > --
> > Philip Ashlock
> > Chief Architect, Data.gov
> > U.S. General Services Administration
> 
> 
> 
> 
> -- 
> Philip Ashlock
> Chief Architect, Data.gov
> U.S. General Services Administration

Received on Wednesday, 11 June 2014 16:21:33 UTC