- From: Phil Archer <phila@w3.org>
- Date: Thu, 18 Aug 2016 14:16:15 +0100
- To: Marcel Fröhlich <marcel.frohlich@gmail.com>, cristiano longo <cristianolongo@gmail.com>
- Cc: "<semantic-web@w3.org>" <semantic-web@w3.org>
Also, you can spend months or years developing a vocabulary, carefully thinking of how you see it being used... and then other people start using it and, well, things tend to change ;-) Which is the topic of an event in Bologna later this year http://event.cwi.nl/drift-a-lod/2016/ (that I'm sorry I can't get to) and, from a process management POV, what my Amsterdam event is also about (https://www.w3.org/2016/11/sdsvoc/). I well remember many years ago DanBri telling me that you can't tell people how to use your spec - people will use it as they want to. So if it suits Cristiano to describe an RSS feed as a dcat:Distribution of a dcat:Dataset that's a website, that's what he'll do. Whether anyone else finds that data useful of course is another matter entirely ;-) Phil. On 18/08/2016 13:36, Marcel Fröhlich wrote: > Hi Cristiano, > > thanks for the clarification. > If RSS feed is summary of the site's content with links then I agree with > Phil's proposal to type it as a distribution. > > Regards, Marcel > > 2016-08-18 14:28 GMT+02:00 cristiano longo <cristianolongo@gmail.com>: > >> Rss is a summary of the web site. It contains titles, summary and links to >> all the articles of the site. Of course, in generale an rss feed may point >> to contents from differenti sites. >> >> Il 18/ago/2016 02:23 PM, "Marcel Fröhlich" <marcel.frohlich@gmail.com> ha >> scritto: >> >>> Hello Phil, >>> >>> is this really the intention of DCAT to allow a dataset denote something >>> different than the distribution? >>> (assuming RSS feed content != other website content) >>> >>> I never checked DCAT standards docs in detail, but my intuitive >>> understanding is, that different distributions of a dataset should have the >>> intention to cover the same content, just differing regarding format, >>> access method and maybe version or specific restrictions. >>> >>> If a distribution is just "some data content" that is part of a dataset, >>> then we look at a collection of of rather arbitrary objects, which is not >>> what I'd like to be the definition of a dataset. >>> >>> If such a concept were required, I'd rather introduce the notion of a >>> "component" type, to make clear that there is additional structure. Whereas >>> in my opinion dataset vs. distribution should be looked at more like >>> intentional definition vs. extensional definition ( >>> https://en.wikipedia.org/wiki/Extensional_and_intensional_definitions). >>> >>> Cheers, Marcel >>> >>> @FroehlichMarcel >>> >>> >>> 2016-08-18 13:22 GMT+02:00 Phil Archer <phila@w3.org>: >>> >>>> >>>> >>>> On 18/08/2016 11:16, cristiano longo wrote: >>>> >>>>> Thanks Phil. I have two use cases. The former is simpler: i have an rdf >>>>> dataset if events and associated a sparql endpoint, an rss feed and a >>>>> Calendar file. Here it is quite clear that the rdf dataset have to be >>>>> modeled as dcat!:Dataset whereas the sparql endpoint, the rss feed and >>>>> the >>>>> iCal file are distributions of this dataset. >>>>> >>>> >>>> Yep. >>>> >>>> The latter case is more >>>> >>>>> confusing: i have just a website with some articles and an associated >>>>> rss >>>>> feed. I see nothing here that is obviously a dataset (however the >>>>> definition of set of data is really large) >>>>> >>>> >>>> Ah, right. Yes, the definition of a dataset is so broad that a website >>>> counts, and I guess therefore an RSS feed counts as a distribution. >>>> >>>> Cheers >>>> >>>> Phil >>>> >>>> >>>>> Il 18/ago/2016 11:27 AM, "Phil Archer" <phila@w3.org> ha scritto: >>>>> >>>>> Hi Cristiano, >>>>>> >>>>>> dcat:Dataset and dcat:Distribution are not disjoint so you can have an >>>>>> RSS >>>>>> feed as an instance of both. But it wouldn't be correct to have a >>>>>> dcat:Dataset that was a collection of RSS feeds and then each of those >>>>>> feeds as a Distribution since the Distribution is a way of accessing >>>>>> the >>>>>> full dataset. You may have an API that allows you to select subsets of >>>>>> the >>>>>> dataset (that's a hot topic for me at the moment) and so your >>>>>> Distribution, >>>>>> which is an API, might well yield a single RSS feed but you'll need to >>>>>> think of it like that. >>>>>> >>>>>> I'm confusing myself just writing this but I hope it makes some sense. >>>>>> And, it would be remiss of me, when talking about DCAT, not to point >>>>>> you to >>>>>> the workshop on that topic later this year: >>>>>> https://www.w3.org/2016/11/sds >>>>>> voc/ Your question is very much in scope for that. >>>>>> >>>>>> Cheers >>>>>> >>>>>> Phil. >>>>>> >>>>>> On 17/08/2016 22:01, Cristiano Longo wrote: >>>>>> >>>>>> Hi folks, I'm creating a list of intresting RSS feeds. I suppose that >>>>>>> thet can be modeled as DCAT datasets (with RSS as distribution), am I >>>>>>> right? >>>>>>> >>>>>>> >>>>>>> >>>>>>> Any suggestion or comment is welcome, >>>>>>> >>>>>>> CL >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> -- >>>>>> >>>>>> >>>>>> Phil Archer >>>>>> W3C Data Activity Lead >>>>>> http://www.w3.org/2013/data/ >>>>>> >>>>>> http://philarcher.org >>>>>> +44 (0)7887 767755 >>>>>> @philarcher1 >>>>>> >>>>>> >>>>> >>>> -- >>>> >>>> >>>> Phil Archer >>>> W3C Data Activity Lead >>>> http://www.w3.org/2013/data/ >>>> >>>> http://philarcher.org >>>> +44 (0)7887 767755 >>>> @philarcher1 >>>> >>>> >>> > -- Phil Archer W3C Data Activity Lead http://www.w3.org/2013/data/ http://philarcher.org +44 (0)7887 767755 @philarcher1
Received on Thursday, 18 August 2016 13:13:48 UTC