- From: Jiri Prochazka <ojirio@gmail.com>
- Date: Thu, 29 Feb 2024 22:24:11 +0100
- To: Semantic Web <semantic-web@w3.org>
- Message-ID: <62f09339-feec-42d6-ab69-d7f060e6e386@gmail.com>
I've been out of the semantic web world for a while and I'm looking at the published drafts of the RDF 1.2 documents. I'm a bit confused and seeking confirmation that my understanding of the problematic, the goals of the RDF 1.2 (regarding this topic) is correct and seeking information if there are some other initiatives (quite possibly non W3C) on the topic. A concrete "triple" can be one of the following things: 1. Triple as the meaning intended to be conveyed by it. 2. Triple as a particular assertion (a statement). Suitable for asserting that some person believes it (or not), asserted it, when, where, how. 3. Triple as an assertion (of a particular triple), a type (or existential variable?) of which point 2. assertions are instances. 4. Triple as a particular (subject, property, object) tuple, as an abstract, mathematical object. Suitable for asserting that it is part of some graph, or relate to an assertion of it (2.). For a graph analogously: 5. Graph as the meaning intended to be conveyed by it. 6. Graph as a particular collections of assertions (statements). Suitable for asserting that some person believes them (or not), asserted them, when, where, how. 7. Graph as an assertion (of a particular graph), a type (or existential variable?) of which point 6. assertions are instances. 8. Graph as a particular set of triples, as an abstract, mathematical object. Suitable for asserting that it is part of some dataset, or relate to an assertion of it (6.). Then there are datasets, as collections of (possibly named) graphs (with possibly a default graph). Point 1. is the most common usage of RDF. Reification [1] allows expressions about a triple as in point 2. (being defined based on the S/P/O IRI referents, not the IRIs). From what I've read, RDF 1.2 aims to introduce a triple as a RDF term to be able to expressions about a triple in a sense of point 3. Both reification and quoted triples don't assert the triple. That can be done easily by just asserting the triple (as in point 1.) as part of the graph (or use the new Turtle annotation syntax {| ... |}). Regarding n-ary relations, there are the established approaches [2] which work well. There's a way of relating between simple 2-ary relations and more detailed relations [3] (a work I've been part of) like these: <#alice> ex:hasRead <#beekeeping> . <#myreadingevent> a ex:ReadingEvent ; ex:reader <#alice> ; ex:book <#beekeeping> ; ex:when "2010-11-01"^^xsd:date . From what I understand, RDF 1.2 doesn't bring anything new regarding n-ary relations. However I've seen examples of RDF 1.2 like this: <#room1> ex:temperature 100 {| ex:scale ex:DegreeCelsius |}. Surely the domain of the ex:scale relation is a measurement (expressing the measurement is in Celsius), not an assertion of triple (expressing the triple is in Celsius makes no sense). Are such examples are errors and should be corrected and discouraged? If not, it seems to me creating a new concept to be used as a domain of relations such as ex:scale which would mean the relation expressed by the triple assertion would complicate things a lot. As I understand it, this is exactly what RDF-star does, introducing the transparency-enabling property [4], but I see no word of it in RDF 1.2 spec drafts. Is this still in heavy discussions? Lastly on topic of datasets, which is the most valuable topic for me, my understanding that they are out of scope of RDF 1.2 and the status is the same as in 2014 note "RDF 1.1: On Semantics of RDF Datasets" [5]. Has there been any work (W3C or not) to have a vocabulary allowing to express that a particular dataset was created with intent of particular semantics as described in the 2014 note? Personally the existing RDF 1.1 ways of expressing n-ary relations and datasets (semantics "Each named graph defines its own context") are the most useful tools to me. Perhaps quoted triples are intended to be a step towards standardizing dataset semantics? Cheers, Jiri Prochazka [1] https://www.w3.org/TR/2014/REC-rdf11-mt-20140225/#reification [2] https://www.w3.org/TR/swbp-n-aryRelations/ [3] https://smiy.sourceforge.net/prv/spec/propertyreification.html [4] https://w3c.github.io/rdf-star/cg-spec/2021-12-17.html#selective-ref-transparency [5] https://www.w3.org/TR/2014/NOTE-rdf11-datasets-20140225/
Received on Thursday, 29 February 2024 21:24:20 UTC