summaryrefslogtreecommitdiff
path: root/word-aligner/aligner.pl
diff options
context:
space:
mode:
authorredpony <redpony@ec762483-ff6d-05da-a07a-a48fb63a330f>2010-10-29 17:14:57 +0000
committerredpony <redpony@ec762483-ff6d-05da-a07a-a48fb63a330f>2010-10-29 17:14:57 +0000
commit1305f38dba367f31754044cabc7b2bea9b8a2074 (patch)
tree0ccfd08b62d4c1f22df2fcace03d0f2335995e15 /word-aligner/aligner.pl
parent78f50e0c3c63de2149045c5afb307e9a3cacff82 (diff)
more wa
git-svn-id: https://ws10smt.googlecode.com/svn/trunk@702 ec762483-ff6d-05da-a07a-a48fb63a330f
Diffstat (limited to 'word-aligner/aligner.pl')
-rwxr-xr-xword-aligner/aligner.pl8
1 files changed, 6 insertions, 2 deletions
diff --git a/word-aligner/aligner.pl b/word-aligner/aligner.pl
index 508dbd8d..f0733449 100755
--- a/word-aligner/aligner.pl
+++ b/word-aligner/aligner.pl
@@ -33,8 +33,7 @@ print STDERR " Using mkcls in: $mkcls\n\n";
die "Don't have an orthographic normalizer for $f_lang\n" unless -f "$SCRIPT_DIR/ortho-norm/$f_lang.pl";
die "Don't have an orthographic normalizer for $e_lang\n" unless -f "$SCRIPT_DIR/ortho-norm/$e_lang.pl";
-my @stages = qw(markov);
-my @directions = qw(f-e e-f);
+my @directions = qw(f-e);
my $corpus = 'c';
@@ -112,8 +111,13 @@ sub make_stage {
open CDEC, ">$stage_dir/cdec.ini" or die "Can't write $stage_dir/cdec.ini: $!";
print CDEC <<EOT;
formalism=lextrans
+lextrans_use_null=true
intersection_strategy=full
+
grammar=$align_dir/grammars/corpus.$direction.lex-grammar.gz
+# grammar=$align_dir/grammars/freq_grammar.$direction.gz
+# per_sentence_grammar_file=$align_dir/grammars/psg.$direction
+
feature_function=LexicalPairIdentity
feature_function=LexicalPairIdentity C $align_dir/grammars/corpus.class.$first $align_dir/grammars/voc2class.$second
feature_function=LexicalPairIdentity S $align_dir/grammars/corpus.stemmed.$first $align_dir/grammars/${second}stem.map