diff --git a/.gitignore b/.gitignore index f1572da..6383c12 100644 --- a/.gitignore +++ b/.gitignore @@ -10,6 +10,7 @@ *.3bo *.mod vocab.* +!vocab.flm .project .classpath bin/ccg2xml.py diff --git a/ccgbank/build.properties b/ccgbank/build.properties old mode 100755 new mode 100644 diff --git a/ccgbank/build.xml b/ccgbank/build.xml old mode 100755 new mode 100644 diff --git a/ccgbank/models/hypertagger/vocab.flm b/ccgbank/models/hypertagger/vocab.flm new file mode 100644 index 0000000..09f9d5b --- /dev/null +++ b/ccgbank/models/hypertagger/vocab.flm @@ -0,0 +1,21 @@ + +## flm file for determining the vocab + +3 + +## word (W) unigram + +W : 0 w.count w.lm 1 + 0 0 + +## supertag (T) unigram + +T : 0 t.count t.lm 1 + 0 0 + +## pos tag (P) unigram + +P : 0 p.count p.lm 1 + 0 0 + + diff --git a/ccgbank/models/parser/vocab.flm b/ccgbank/models/parser/vocab.flm new file mode 100644 index 0000000..4de66c2 --- /dev/null +++ b/ccgbank/models/parser/vocab.flm @@ -0,0 +1,71 @@ + +## flm file for determining the vocab + +13 + +## headword (W) unigram + +W : 0 w.count w.lm 1 + 0 0 + +## headword top (WT) + +WT : 0 wt.count wt.lm 1 + 0 0 + +## headword sibling (WS) + +WS : 0 ws.count ws.lm 1 + 0 0 + +## expansion (E) + +E : 0 e.count e.lm 1 + 0 0 + +## head (H) cat + +H : 0 h.count h.lm 1 + 0 0 + +## parent (P) cat + +P : 0 p.count p.lm 1 + 0 0 + +## lexcat parent (CP) + +CP : 0 cp.count cp.lm 1 + 0 0 + +## lexcat top (CT) + +CT : 0 ct.count ct.lm 1 + 0 0 + +## head postag top (TT) + +TT : 0 tt.count tt.lm 1 + 0 0 + +## head postag parent (T) + +T : 0 t.count t.lm 1 + 0 0 + +## sibling (S) cat + +S : 0 s.count s.lm 1 + 0 0 + +## lexcat sibling (CS) + +CS : 0 cs.count cs.lm 1 + 0 0 + +## head postag sibling (TS) + +TS : 0 ts.count ts.lm 1 + 0 0 + + diff --git a/ccgbank/models/supertagger/st.noprior.config b/ccgbank/models/supertagger/st.noprior.config old mode 100755 new mode 100644 diff --git a/ccgbank/models/supertagger/vocab.flm b/ccgbank/models/supertagger/vocab.flm new file mode 100644 index 0000000..09f9d5b --- /dev/null +++ b/ccgbank/models/supertagger/vocab.flm @@ -0,0 +1,21 @@ + +## flm file for determining the vocab + +3 + +## word (W) unigram + +W : 0 w.count w.lm 1 + 0 0 + +## supertag (T) unigram + +T : 0 t.count t.lm 1 + 0 0 + +## pos tag (P) unigram + +P : 0 p.count p.lm 1 + 0 0 + + diff --git a/ccgbank/original/models/supertagger/vocab.flm b/ccgbank/original/models/supertagger/vocab.flm new file mode 100644 index 0000000..09f9d5b --- /dev/null +++ b/ccgbank/original/models/supertagger/vocab.flm @@ -0,0 +1,21 @@ + +## flm file for determining the vocab + +3 + +## word (W) unigram + +W : 0 w.count w.lm 1 + 0 0 + +## supertag (T) unigram + +T : 0 t.count t.lm 1 + 0 0 + +## pos tag (P) unigram + +P : 0 p.count p.lm 1 + 0 0 + +