[developers] Punctuation and "-default-les" type mapping in PET/ERG

Stephan Oepen oe at ifi.uio.no
Sun Apr 13 15:53:29 CEST 2008

hi again,

> > Instead, I think it is finally time (for someone) to fix this hack in
> > the handling of generic entries in PET, so the PRED value for an unknown
> > word is a string which is constructed on the fly from its stem orthography
> > and its POS, using the following template:
> >
> >  "_ORTH_POS_unk_rel"
> I would agree with Dan's proposed solution. From what I can see, the
> necessary change in PET is relatively straightforward (basically
> related to the function lex_parser::add_surface_mod()). I will see if
> I can find time to make an initial attempt on this soon.

this may still be a useful intermediate mechanism, but it would further
grow our inventory of procedural elements of FS frobbing inside of PET
(and the LKB, in principle).  at a fairly recent MRS-related meeting, a
proposal emerged that will simplify PRED manufacturing for generic LEs,
see `PRED with Internal Structure' at:


in terms of MRS read-out from feature structures, this will be trivial
to implement (i volunteered for the LKB side).  for generic entries, it
leaves the question of how the LE can get hold of the surface form (in
the case of no external morphology or stemming, it would have to be the
inflected form, as the ERG at least opts to consider generic entries as
inflected in the lexicon, i.e. words) and PoS tag.  again, long-term i
expect the strategy sketched in my earlier message today can take care
of this.  

short-term, i am wondering whether we should back-port the ability for
lexical entries to see token IFSs as arguments to the main PET branch.
peter, is that bit of the new functionality sufficiently isolated, so
as to make it feasible to back-port it relatively quickly?

                                                          best  -  oe

+++ Universitetet i Oslo (IFI); Boks 1080 Blindern; 0316 Oslo; (+47) 2284 0125
+++     CSLI Stanford; Ventura Hall; Stanford, CA 94305; (+1 650) 723 0515
+++       --- oe at ifi.uio.no; oe at csli.stanford.edu; stephan at oepen.net ---

More information about the developers mailing list