[developers] yy mode problem

Francis Bond fcbond at gmail.com
Wed Nov 15 02:17:59 CET 2006


G'day,

I
> I used yy mode as input to PET to run my experiment last night, but the
> output is quite strange.  Below is the detailed info:
> ==========================================
> Command line: cat try.yy | cheap -yy -tok=yy -packing -result=1 -default-les
> /(path)/english.grm 2> try.parse

> ===========================================
> output:
> reading `/lingo/delphin/erg/pet/english.set'... including
> `/lingo/delphin/erg/pet/common.set'...
> `/lingo/delphin/erg/pet/global.set'...
> loading `/lingo/delphin/erg/english.grm' (LinGO (Jan-06))
> reading ME model `/lingo/delphin/erg/rondane.mem'... [11846
> features]
> 58789 types in 3.7 s
> (1) `russian' [0] --- 0 (-0.00|0.00s) <3:15> (65.4K) [0.0s]
> warning: unknown type `$generic_mass_count_noun' in POS mapping
> warning: unknown type `$generic_mass_count_noun' in POS mapping

It looks as though the ERG is defining a mapping to a lexical type in
english.set that doesn't exist in the grammar english.grm.  Either you
are using files from different versions, or Dan forgot to update
english.set.    If you can find out what the generic_mass_count_noun
should be called (e.g. by dumping the lexicon from cheap) then you
should be able to fix it yourself in english.set.  There is (some)
documentation of the mappings at:
http://wiki.delph-in.net/moin/PetInput.

-- 
Francis Bond  <www.kecl.ntt.co.jp/icl/mtg/members/bond/>
NTT Communication Science Laboratories | Natural Language Research Group



More information about the developers mailing list