- From: Richard Cyganiak <richard@cyganiak.de>
- Date: Sat, 20 Dec 2014 00:08:10 +0000
- To: Cristiano Longo <longo@dmi.unict.it>
- Cc: SW-forum Web <semantic-web@w3.org>, Fadi Maali <fadi.maali@deri.org>, "<vassilios.peristeras@deri.org>" <vassilios.peristeras@deri.org>
Hi Cristiano, > On 18 Dec 2014, at 22:58, Cristiano Longo <longo@dmi.unict.it> wrote: >>> Now, let us suppose that I have the same dataset reported in two different catalogues, which reports different (not necessarily contraddictory) information. Should I have a single dataset instance and two catalog records, or two dataset instances as well? >> >> “Instance” is not a terribly useful concept in RDF. This isn't OO. >> >> There *is* just a single dataset anyway, according to your supposition. The question is just whether you refer to that dataset by a single name (IRI) in both catalogues, or by two different names (where asserting owl:sameAs between them would be correct). There is nothing wrong with either approach, and which one to choose depends on various practical considerations. > > I borrowed the term "instance" from Description Logic. My issue is about merging two catalogs with two different entries, one for each catalog, about the same dataset, where different means which reports different information. Let us consider for example two catalogues, say C1 and C2, and a dataset D with title "the title" and a landing page http://landingpage.org. Of course we have just one data set, but two different (in the real world we are describing) entries e1 and e2, one for each catalog. > > Now, let us consider the case in which, may be for a typo, the title of D reported in C2 is wrong. Can I model this in DCAT? I'm not sure but examining the example in the documentation I suppose that CatalogRecord and hasCatalogRecord are not suitable for this purpose. 1. A single thing can have multiple IRIs referring to it. So, D can be identified by different IRIs in the to catalogs. 2. Even if D was identified by the same IRI in both catalogs, a consumer of the catalogs may use named graphs to keep the data from different sources separate. This way, they can query the joint dataset, but where necessary still work out what triple came from where. Hope that helps, Richard
Received on Saturday, 20 December 2014 00:08:38 UTC