You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The idea is that one could build-up a full DMRS from smaller components by linking "slots", or unused arguments, to the index, top, or xarg of other DMRS components. These slots could be open-ended (e.g., linking FAKEARG/NEQ would not raise an error) or constrained by a SEM-I. Errors would be raised if, e.g,, a user tried to draw a /EQ link to something that must be at a different scope (because of other /H links), or if the link type is not compatible with the start and end nodes (based on cvarsort, predicate pos, etc.).
The composition (like a derivation) of a DMRS could maybe be stored and accessed. It would be really cool if the composition could be extracted somehow from derivation trees (but it would probably require some kind of mapping from rule names to composition operations).
It would also be neat if the composition functions could be abstracted over the relevant operations on DMRS, MRS, EDS, etc., so that we could also model, e.g., MRS composition with the same API.
The text was updated successfully, but these errors were encountered:
See http://moin.delph-in.net/StanfordAlgebraAdditions
The idea is that one could build-up a full DMRS from smaller components by linking "slots", or unused arguments, to the
index
,top
, orxarg
of other DMRS components. These slots could be open-ended (e.g., linkingFAKEARG/NEQ
would not raise an error) or constrained by a SEM-I. Errors would be raised if, e.g,, a user tried to draw a/EQ
link to something that must be at a different scope (because of other/H
links), or if the link type is not compatible with the start and end nodes (based on cvarsort, predicate pos, etc.).The composition (like a derivation) of a DMRS could maybe be stored and accessed. It would be really cool if the composition could be extracted somehow from derivation trees (but it would probably require some kind of mapping from rule names to composition operations).
It would also be neat if the composition functions could be abstracted over the relevant operations on DMRS, MRS, EDS, etc., so that we could also model, e.g., MRS composition with the same API.
The text was updated successfully, but these errors were encountered: