GitHub/gf-core
1
0
Fork 1
mirror of https://github.com/GrammaticalFramework/gf-core.git synced 2026-04-09 04:59:31 -06:00
Code Issues Packages Projects Releases Wiki Activity

just for fun - grammar for trigram models in GF

Browse Source
This commit is contained in:
krasimir
2009-10-08 13:13:48 +00:00
parent 30638abe98
commit d920a0c53d
4 changed files with 153 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

84
examples/trigram/Shaw.gf Normal file
View File

@@ -0,0 +1,84 @@
abstract Shaw = Trigram ** {
-- This module contains Trigram model of this quote from George Bernard Shaw:
--
-- The reasonable man adapts himself to the world; the unreasonable one persists in trying
-- to adapt the world to himself. Therefore all progress depends on the unreasonable man.
data
the_W,reasonable_W,man_W,adapts_W,himself_W,to_W,world_W,unreasonable_W,
one_W,persists_W,in_W,trying_W,adapt_W,therefore_W,all_W,progress_W,depends_W,on_W : Word ;
the_U : Unigram the_W ; --# prob 0.179
reasonable_U : Unigram reasonable_W ; --# prob 0.036
man_U : Unigram man_W ; --# prob 0.071
adapts_U : Unigram adapts_W ; --# prob 0.036
himself_U : Unigram himself_W ; --# prob 0.071
to_U : Unigram to_W ; --# prob 0.107
world_U : Unigram world_W ; --# prob 0.071
unreasonable_U : Unigram unreasonable_W ; --# prob 0.071
one_U : Unigram one_W ; --# prob 0.036
persists_U : Unigram persists_W ; --# prob 0.036
in_U : Unigram in_W ; --# prob 0.036
trying_U : Unigram trying_W ; --# prob 0.036
adapt_U : Unigram adapt_W ; --# prob 0.036
therefore_U : Unigram therefore_W ; --# prob 0.036
all_U : Unigram all_W ; --# prob 0.036
progress_U : Unigram progress_W ; --# prob 0.036
depends_U : Unigram depends_W ; --# prob 0.036
on_U : Unigram on_W ; --# prob 0.036
the_reasonable_B : Bigram the_W reasonable_W ; --# prob 0.037
reasonable_man_B : Bigram reasonable_W man_W ; --# prob 0.037
man_adapts_B : Bigram man_W adapts_W ; --# prob 0.037
adapts_himself_B : Bigram adapts_W himself_W ; --# prob 0.037
himself_to_B : Bigram himself_W to_W ; --# prob 0.037
to_the_B : Bigram to_W the_W ; --# prob 0.037
the_world_B : Bigram the_W world_W ; --# prob 0.037
world_the_B : Bigram world_W the_W ; --# prob 0.037
the_unreasonable_B : Bigram unreasonable_W one_W ; --# prob 0.074
unreasonable_one_B : Bigram unreasonable_W one_W ; --# prob 0.037
one_persists_B : Bigram one_W persists_W ; --# prob 0.037
persists_in_B : Bigram persists_W in_W ; --# prob 0.037
in_trying_B : Bigram in_W trying_W ; --# prob 0.037
trying_to_B : Bigram trying_W to_W ; --# prob 0.037
to_adapt_B : Bigram to_W adapt_W ; --# prob 0.037
adapt_the_B : Bigram adapt_W the_W ; --# prob 0.037
the_world_B : Bigram the_W world_W ; --# prob 0.037
world_to_B : Bigram world_W to_W ; --# prob 0.037
to_himself_B : Bigram to_W himself_W ; --# prob 0.037
himself_therefore_B : Bigram himself_W therefore_W ; --# prob 0.037
therefore_all_B : Bigram therefore_W all_W ; --# prob 0.037
all_progress_B : Bigram all_W progress_W ; --# prob 0.037
progress_depends_B : Bigram progress_W depends_W ; --# prob 0.037
depends_on_B : Bigram depends_W on_W ; --# prob 0.037
on_the_B : Bigram on_W the_W ; --# prob 0.037
unreasonable_man_B : Bigram unreasonable_W man_W ; --# prob 0.037
the_reasonable_man_T : Trigram the_W reasonable_W man_W ; --# prob 0.038
reasonable_man_adapts_T : Trigram reasonable_W man_W adapts_W ; --# prob 0.038
man_adapts_himself_T : Trigram man_W adapts_W himself_W ; --# prob 0.038
adapts_himself_to_T : Trigram adapts_W himself_W to_W ; --# prob 0.038
himself_to_the_T : Trigram himself_W to_W the_W ; --# prob 0.038
to_the_world_T : Trigram to_W the_W world_W ; --# prob 0.038
the_world_the_T : Trigram the_W world_W the_W ; --# prob 0.038
world_the_unreasonable_T : Trigram world_W the_W unreasonable_W ; --# prob 0.038
the_unreasonable_one_T : Trigram the_W unreasonable_W one_W ; --# prob 0.038
unreasonable_one_persists_T : Trigram unreasonable_W one_W persists_W ; --# prob 0.038
one_persists_in_T : Trigram one_W persists_W in_W ; --# prob 0.038
persists_in_trying_T : Trigram persists_W in_W trying_W ; --# prob 0.038
in_trying_to_T : Trigram in_W trying_W to_W ; --# prob 0.038
trying_to_adapt_T : Trigram trying_W to_W adapt_W ; --# prob 0.038
to_adapt_the_T : Trigram to_W adapt_W the_W ; --# prob 0.038
adapt_the_world_T : Trigram adapt_W the_W world_W ; --# prob 0.038
the_world_to_T : Trigram the_W world_W to_W ; --# prob 0.038
world_to_himself_T : Trigram world_W to_W himself_W ; --# prob 0.038
to_himself_therefore_T : Trigram to_W himself_W therefore_W ; --# prob 0.038
himself_therefore_all_T : Trigram himself_W therefore_W all_W ; --# prob 0.038
therefore_all_progress_T : Trigram therefore_W all_W progress_W ; --# prob 0.038
all_progress_depends_T : Trigram all_W progress_W depends_W ; --# prob 0.038
progress_depends_on_T : Trigram progress_W depends_W on_W ; --# prob 0.038
depends_on_the_T : Trigram depends_W on_W the_W ; --# prob 0.038
on_the_unreasonable_T : Trigram on_W the_W unreasonable_W ; --# prob 0.038
the_unreasonable_man_T : Trigram the_W unreasonable_W man_W ; --# prob 0.038
}

23
examples/trigram/ShawCnc.gf Normal file
View File

@@ -0,0 +1,23 @@
concrete ShawCnc of Shaw = TrigramCnc ** {
lin
the_W = "the" ;
reasonable_W = "reasonable" ;
man_W = "man" ;
adapts_W = "adapts" ;
himself_W = "himself" ;
to_W = "to" ;
world_W = "world" ;
unreasonable_W = "unreasonable" ;
one_W = "one" ;
persists_W = "persists" ;
in_W = "in" ;
trying_W = "trying" ;
adapt_W = "adapt" ;
therefore_W = "therefore" ;
all_W = "all" ;
progress_W = "progress" ;
depends_W = "depends" ;
on_W = "on" ;
}

34
examples/trigram/Trigram.gf Normal file
View File

@@ -0,0 +1,34 @@
abstract Trigram = {
cat
-- A lexicon is a set of 'Word's
Word ;
-- All N-gram instances seen in the corpus are abstract syntax constants
Unigram (a : Word) ;
Bigram (a,b : Word) ;
Trigram (a,b,c : Word) ;
-- A text is a sequence words where the sequence is indexed by the last two tokens
Seq (a,b : Word) ;
-- The estimated probability of the trigram 'a b c' is the total probability of all
-- trees of type Prob a b c.
Prob (a,b,c : Word) ;
data
-- Here we construct sequence by using nil and cons. The Prob argument ensures
-- that the sequence contains only valid N-grams and contributes with the right
-- probability mass
nil : (a,b,c : Word) -> Prob a b c -> Seq b c ;
cons : ({a,b} : Word) -> Seq a b -> (c : Word) -> Prob a b c -> Seq b c ;
-- Here we construct probabilities. There are two ways: by trigrams, by bigrams and
-- by unigrams. Since the trigramP, bigramP, unigramP functions have some associated
-- probabilities as well this results in linear smoothing between the unigram, bigram
-- and trigram models
trigramP : ({a,b,c} : Word) -> Trigram a b c -> Prob a b c ;
bigramP : ({a,b,c} : Word) -> Bigram a b -> Bigram b c -> Prob a b c ;
unigramP : ({a,b,c} : Word) -> Unigram a -> Unigram b -> Unigram c -> Prob a b c ;
}

12
examples/trigram/TrigramCnc.gf Normal file
View File

@@ -0,0 +1,12 @@
concrete TrigramCnc of Trigram = {
lincat
Word,Seq = Str;
Unigram, Bigram, Trigram, Prob = {} ;
lin
nil a b c _ = a ++ b ++ c ;
cons _ _ l c _ = l ++ c ;
}