| << Prev | - Up - | Next >> |
Assuming the above trees represent the two possible meanings of Every yogi has a guru, we can use the following dominance constraint to describe both trees:
apply apply
' ` ' `
' ` ' `
apply lambda(2) apply lambda(1)
' ` ! ' ` !
' ` ! ' ` !
a guru . every yogi .
. .
. .
. .
. .
. .
..
apply
' `
' `
apply var(1)
' `
' `
has var(2)
where the dotted lines indicate dominance and the other lines immediate dominance. If the lambda(2) node dominates the every yogi-tree, we get the second reading and if the every yogi-tree dominates the a guru-tree, we get the first reading.
This is the intuition. We now see more precisely what the constraint language is and how a constraint solver can be implemented which given a dominance logic description, yields the set of trees satisfying that description.
| << Prev | - Up - | Next >> |