search

w3c / rdf-star

RDF-star specification
https://w3c.github.io/rdf-star/
Other
118 stars 23 forks source link

RDF-star semantics (as currently defined) is non-monotonic #262

Open pchampin opened 2 years ago

pchampin commented 2 years ago

... and we obviously do not want this!

What is the problem?

More precisely:

<< :s :p "foo"^^xsd:integer >> :accordingTo :pa.

entails

<< s: :p [] >> :accordingTo :pa.

under simple entailment, but does not entail that anymore under D-entailment recognizing xsd:integer.

How to fix it?

An easy fix would be to change step 2.4 of the unstar mapping, more precisely:

  • (b, unstar:object, o) unless o is an ill-typed literal

would simply become

  • (b, unstar:object, o)

It means that ill-formed literals in quoted triples would now make the whole graph inconsistent -- which the current semantics aimed to avoid, but clearly that was not such a good idea...

hartig commented 2 years ago

Good catch! I am fine with the proposed fix.

TallTed commented 2 years ago

Probably best to fix the xsd;integer typo above, making it xsd:integer, in the "More precisely" discussion.

pchampin commented 2 years ago

Probably best to fix the xsd;integer typo above, making it xsd:integer, in the "More precisely" discussion.

done, thanks

afs commented 2 years ago

+1 to the proposed fix.