diff options
author | redpony <redpony@ec762483-ff6d-05da-a07a-a48fb63a330f> | 2010-10-28 00:22:42 +0000 |
---|---|---|
committer | redpony <redpony@ec762483-ff6d-05da-a07a-a48fb63a330f> | 2010-10-28 00:22:42 +0000 |
commit | 22e05bf807ad59bfad38fcdf35bb51524034e23b (patch) | |
tree | ce537739d375d5216a9558414dd9d8dce065f6fc /word-aligner/aligner.pl | |
parent | ecde8cb600b24c31b062f8f53d57641e3fa23379 (diff) |
change stem handling
git-svn-id: https://ws10smt.googlecode.com/svn/trunk@693 ec762483-ff6d-05da-a07a-a48fb63a330f
Diffstat (limited to 'word-aligner/aligner.pl')
-rwxr-xr-x | word-aligner/aligner.pl | 4 |
1 files changed, 4 insertions, 0 deletions
diff --git a/word-aligner/aligner.pl b/word-aligner/aligner.pl index e23c2beb..508dbd8d 100755 --- a/word-aligner/aligner.pl +++ b/word-aligner/aligner.pl @@ -105,7 +105,9 @@ sub make_stage { my ($direction) = @_; my $stage_dir = "$align_dir/model-$direction"; my $first = $direction; + my $second = $direction; $first =~ s/^(.+)-.*$/$1/; + $second =~ s/^.+-(.+)$/$1/; mkdir $stage_dir; open CDEC, ">$stage_dir/cdec.ini" or die "Can't write $stage_dir/cdec.ini: $!"; print CDEC <<EOT; @@ -113,6 +115,8 @@ formalism=lextrans intersection_strategy=full grammar=$align_dir/grammars/corpus.$direction.lex-grammar.gz 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 feature_function=InputIdentity feature_function=OutputIdentity feature_function=RelativeSentencePosition $align_dir/grammars/corpus.class.$first |