[Application-profiles-ig] Difference between profile and vocabulary

Tue Mar 3 16:19:07 GMT 2020

Hi All, I accept the decision to strip namespaces in a simple profile, but I still couldn't find it logical as an implementer. The most known definition of Application Profile is: schemas consist of data elements drawn from one or more namespaces, combined, and optimized for a local application[7]. This 'mix and match' is one of the best use cases of an application profile. As long as it is neither a QName or URI [10], then it is a vocabulary that too non-linkable. My question is, do we still need to walk all the way back to the beginning of the semantic web? The QName concept [1] is from the XML namespace definition [2], and it is only valid of the prefixes are declared locally. The properties are pointers to concepts even if the database is local, that is the main reason we need them as URIs [11]. Also, we use QNames, not just for properties; For example, from the prototypes [3], 'xsd:year' is an example we use for constraints. There is no universal notion about prefixes, and that will make the implementation difficult if the prefixes are not declared. An easy workaround for us to use valid QNames without altering the current CSV structure is to follow any prefix consensus like prefix.cc [4] or LOV [8]. Since CSV expression is technical, we should define the format explicitly to make the implementation easier. Something like: "URIs can be either QNames with prefixes from prefix.cc or RFC3986 IRIs" [5]. Another good thing about such a declaration is, the implementers can download all prefix lists to create the DSP parser[6] or use an API like LOV [9]. The more standards we break, the more complex it to implement and less acceptance for our proposals. Nishad [1]https://www.w3.org/2001/tag/doc/qnameids[2]https://www.w3.org/TR/xml-names/[3]https://github.com/dcmi/dcap/blob/master/prototypes/simpleFromHackathon/profile2Instance1.csv[4]http://prefix.cc/[5]https://tools.ietf.org/html/rfc3986[6]http://prefix.cc/popular/all.file.csv[7]http://www.ariadne.ac.uk/issue/25/app-profiles/[8]https://lov.linkeddata.es/dataset/lov/vocabs[9]https://lov.linkeddata.es/dataset/lov/api[10]https://www.w3.org/TR/cooluris/[11]https://www.w3.org/Provider/Style/URIOn 2020/03/04 0:53, "Application-profiles-ig on behalf of Thomas Baker" <application-profiles-ig-bounces at lists.www.voudr.comon behalf oftom at tombaker.org> wrote: Dear all, I took an action [1] to follow up here on an interesting question posed by John. On our short call, we agreed that the most minimal profile could consist simply of a list of properties. In other words, the following could be a profile:http://purl.org/dc/terms/creatorhttp://purl.org/dc/terms/titlehttp://purl.org/dc/terms/dateBut because this group had earlier decided that property URIs are not actually a requirement for profiles, the simplest, most minimal profile possible could even just be: creator title date ...without necessarily even having a separate "property label". John asked: If this is so, then what is the essential difference between a profile and a vocabulary? Tom [1]https://hackmd.io/pTAlXlcTQFmyrCutEhisCg?both-- Tom Baker <tom at tombaker.org>