[developers] Sweaglesw ERG binaries don't include SEM-I?

Dan Flickinger danf at stanford.edu
Tue Mar 14 22:55:34 CET 2017


I'll bring the trunk version up to date on the SEM-I front by the end of this week, but will have to discuss with Stephan the merits and drawbacks of making any changes to 1214 at this point.


 Dan

________________________________
From: fcbond at gmail.com <fcbond at gmail.com> on behalf of Francis Bond <bond at ieee.org>
Sent: Tuesday, March 14, 2017 2:31 PM
To: Woodley Packard
Cc: Michael Wayne Goodman; Stephan Oepen; Dan Flickinger; Emily M. Bender; developers at delph-in.net
Subject: Re: Sweaglesw ERG binaries don't include SEM-I?

G'day,

moving this to developers.

In summary:
A side effect of the move to the SEM-I hierarchy for generation is that some of the predicates that were visible for generation (and were in the hand-written erg/etc/erg.smi) did not make it to the new semi).

Many of these were made by Dan to handle input for machine translation from Japanese.

Probably the most used is  def_udef_a_q_rel, but I also use unspec_loc_rel, poss_rel and can_able rel.

  ; supertype of def_q, udef_q, _a_q, some-any_q
  def_udef_some_a_q_rel : ARG0 x, BODY h, RSTR h.
  ; supertype of def_q, udef_q, _a_q
  def_udef_a_q_rel : ARG0 x, BODY h, RSTR h.

  def_q_rel : ARG0 x, BODY h, RSTR h.
  idiom_q_rel : ARG0 x, BODY h, RSTR h.

  ; temp or nontemp 'in, on, at' plus loc_nonsp_rel
  unspec_loc_rel : ARG0 i, ARG1 u, ARG2 u.
  ; nonspecific locative: temp or nontemp
  loc_nonsp_rel : ARG0 i, ARG1 u, ARG2 u.
  ; temp or nontemp 'in, on, at'
  loc_sp_rel : ARG0 i, ARG1 u, ARG2 u.
  ; nontemp 'in, on, at' plus loc_nonsp_rel
  nontemp_sp_or_nonsp_rel : ARG0 i, ARG1 u, ARG2 u.
  ; nontemp 'in, on, at'
  nontemp_loc_sp_rel : ARG0 i, ARG1 u, ARG2 u.
  ; temp 'in, on, at' plus loc_nonsp_rel
  temp_sp_or_nonsp_rel : ARG0 i, ARG1 u, ARG2 u.
  ; temp 'in, on, at'
  temp_loc_sp_rel : ARG0 i, ARG1 u, ARG2 u.

  ; 'can' or 'be able to'
  can_able_rel : ARG0 e, ARG1 h.
  ; compound or 'of'
  compound_or_prep_rel : ARG0 i, ARG1 u, ARG2 u.

It would be great if these could be snuck into the 1214 release.

--
Francis Bond <http://www3.ntu.edu.sg/home/fcbond/>
Francis Bond's Humble Home Page<http://www3.ntu.edu.sg/home/fcbond/>
www3.ntu.edu.sg
Francis Bond. Associate Professor in the Computational Linguistics Lab, Division of Linguistics and Multilingual Studies, Nanyang Technological ...


Division of Linguistics and Multilingual Studies
Nanyang Technological University
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.delph-in.net/archives/developers/attachments/20170314/b869344f/attachment.html>


More information about the developers mailing list