Re: [port-peer-review] reviews
> [4 judges, approver: 3 laws] is ok. (01)
Actually, that was a non-sense.
[4 judges, approver of: 3 laws] was what I meant to write.
I (temporarily) woke up at 4am last night because of this missing
keyword "of" (which, in FCG, reverses the relation direction)! (02)
However, this illustrates the importance of semantic checks
and of using more precise relations to permit such checks.
WebKB-2 does not accept the above uses of the type #approver
because the type #judge is neither a subtype nor a supertype
of #approver. It accepts the more precise formulation:
[4 judges, agent of: (an approval, object: 3 laws)]
and would detect a missing or an additional "of" since
the signature of pm#agent or pm#object would be violated. (03)
This second formulation also permits graph matching on the
concept node of type #approval, or other quantifications
in this concept node, e.g.
[4 judges, agent of: (2 approvals, object: 3 laws)] (04)
Philippe (05)