Re: Overview of options

> On 20. Oct 2023, at 22:49, Peter F. Patel-Schneider <pfpschneider@gmail.com> wrote:
> 
> [My problems in accessing the document appear to have been transitory.]
> 
> There are several variants of opacity, which is not reflected in the table, including opacity of blank nodes, opacity of literals, opacity of IRIs, and opacity of graphs.

1)
I had assumed that the Semantics TF has distilled all the options for different kinds of opacity into three different behaviours [0] that it deemed the most promising and practically relevant. Or is essentially every possible combination still on the table?

2)
I also assume that most proposals define some special syntax, and if you have a special syntax you can define its semantics at will, so back to point 1… The nested graph proposal for example provides a syntax for unasserted referentially transparent statements, but one could of course define the semantics of that syntax differently. It also provides a means to define any semantics of your liking via a respective property and an RDF graph literal datatype, but any other proposal can add that facility as well. 

3)
IMO the real discussion here is if it is better to handle specific requirements on semantics through such an extensible mechanism, or if we can/should settle for only as many semantics as we can usefully differentiate (or that we can find useful combinations of brackets for)? Or a combination of both.
The nested graph proposal does the latter: it figures that unasserted and referentially fully transparent statements are quite popular in practice (i.e. how most people mis-understand the RDF standard reification vocabulary) and reserves the {" … "} syntax for them. It further provides a means to include RDF data with any kind of non-standard semantics and it offers (in the long version of the proposal at [1]) a few pre-configured semantics profiles, and a means to construct your own DIY semantics - so: useful default, maximum extensibility.

> It would be useful to add in lines for labelled property graphs and Wikidata and other formalisms that support quotation or term formation from assertions


What are "opaque literals"?
What does "term formation from assertions" mean?

Thomas


> (and maybe modal logic).
> 
> peter
> 


[0] https://github.com/w3c/rdf-star-wg/wiki/Semantics%3A-Behaviour-catalogue#example-cases-in-concrete-syntax
[1] https://gist.github.com/rat10/eaa109ab56b4d77d29e3a826291f8e72#configurable-semantics

Received on Saturday, 21 October 2023 10:32:59 UTC