SemiRfc

SEM-I

A SEM-I, or SEMantic-Interface is a description of the semantic structures output by the grammar, and may include a number of sections:

• Hierarchical relations among variable sorts and properties
• A list of valid roles and allowed types for their values
• A predicate inventory including the allowed roles for the predicate and further constraints on role values and properties
• (aliases?)
• A list of variable properties and their allowed values
• (ges?)
• (types?)

SEM-Is can be useful for validating the semantic output of grammars without having to load the entire grammar.

Although the variable property mapping(s) (VPM) defined by a grammar are not technically part of the SEM-I, there is a strong connection, nevertheless, seeing as the default VPM should map into the property names and values defined in the SEM-I.

.smi file syntax

The .smi files (e.g. erg.smi, hierarchy.smi etc.) use a simplified (non-TDL) syntax to characterize notions of inheritance (e.g. specializations of predicates) and appropriateness (e.g. the frame of arguments and associated value constraints associated with each predicate). Here's a descriptive example:

   1 ; comments begin with semicolons
   2 
   3 signature:  ; sections begin at column 0 and are followed by a colon
   4   ; definitions (by convention) are indented
   5   i < u.  ; definition syntax depends on the section
   6   ;...
   7 
   8 include: core.smi  ; external files can be included
   9 ; sections in included files are merged with sections in the main file

Proposal based on discussion at the Abbey on 2013-08-28

Woodley argues that the SEM-I should encode the hierarchy of predicates. This will allow a tool to calculate MRS subsumption with only the SEM-I and the MRSs (i.e. without having to parse the grammar). E.g. currently we have:

def_q_rel : ARG0 x, BODY h, RSTR h.

And we'd like something like:

def_q_rel (def_udef_a_q_rel) : ARG0 x, BODY h, RSTR h.

There is also talk of linking preds that differ by sense (e.g. number of arguments, like "he ate" vs "he ate a banana"), or mass/count distinctions ("every paper" vs "all the paper"). This is not trying to recreate something like WordNet.