Open mcodescu opened 8 years ago
I would prefer the syntax proposed in #1626, because it is shorter and re-uses networks better. I agree that we must restrict ourselves to relational networks (i.e. consisting of alignments only). But this could also be imposed as a static semantic condition for #1626. In this light, do you see any advantage of your notation over #1626?
At the moment, networks don't know their elements, all we store is the diagram of the network. If we write extract N
we would have to check that N
contains only alignments or sub-networks containing only alignments. Yes, one can implement such a check and use extract N
.
Given a list of ontologies and a list of alignments between them, we want to extract from each ontology
O
a moduleM
of the signature generated by all symbols ofO
that appear in the correspondences of the alignments. Thus, the alignments involvingO
remain valid forM
, and we can build a network whose elements are the modules and the alignments.I was thinking to add modules of networks as a new constructor for networks, using the syntax
modules of O1, ..., On for A1, ...Am
. The notion does not seem to make sense for other mappings than alignments, soAi
must be alignments.