diff options
Diffstat (limited to 'word-aligner')
| -rw-r--r-- | word-aligner/makefiles/makefile.grammars | 18 | ||||
| -rwxr-xr-x | word-aligner/support/generate_word_pair_features.pl | 11 |
2 files changed, 14 insertions, 15 deletions
diff --git a/word-aligner/makefiles/makefile.grammars b/word-aligner/makefiles/makefile.grammars index 60417ec5..be0644df 100644 --- a/word-aligner/makefiles/makefile.grammars +++ b/word-aligner/makefiles/makefile.grammars @@ -54,28 +54,28 @@ voc2class.e: corpus.e $(MKCLS) voc2class.f: corpus.f $(MKCLS) $(MKCLS) -c$(NCLASSES) -n10 -pcorpus.f -Vvoc2class.f opt -corpus.class.e: corpus.e voc2class.e $(CLASSIFY) +corpus.class.e: corpus.e voc2class.e $(CLASSIFY) voc2class.e corpus.e > $@ -corpus.class.f: corpus.f voc2class.f $(CLASSIFY) +corpus.class.f: corpus.f voc2class.f $(CLASSIFY) voc2class.f corpus.f > $@ -corpus.f-e: corpus.f corpus.e $(MERGE_CORPUS) +corpus.f-e: corpus.f corpus.e $(MERGE_CORPUS) corpus.f corpus.e > $@ -corpus.e-f: corpus.f corpus.e $(MERGE_CORPUS) +corpus.e-f: corpus.f corpus.e $(MERGE_CORPUS) corpus.e corpus.f > $@ -corpus.f-e.model1: corpus.f-e $(MODEL1) - $(MODEL1) -v -V corpus.f-e > $@ +corpus.f-e.model1: corpus.f-e + $(MODEL1) -v corpus.f-e > $@ -corpus.e-f.model1: corpus.e-f $(MODEL1) +corpus.e-f.model1: corpus.e-f $(MODEL1) -v -V corpus.e-f > $@ -corpus.f-e.full-model1: corpus.f-e $(MODEL1) +corpus.f-e.full-model1: corpus.f-e $(MODEL1) -t -999999 -v -V corpus.f-e > $@ -corpus.e-f.full-model1: corpus.e-f $(MODEL1) +corpus.e-f.full-model1: corpus.e-f $(MODEL1) -t -999999 -v -V corpus.e-f > $@ corpus.f-e.lex-grammar.gz: corpus.f-e corpus.f-e.model1 corpus.e-f.model1 diff --git a/word-aligner/support/generate_word_pair_features.pl b/word-aligner/support/generate_word_pair_features.pl index b722ee49..b28f6feb 100755 --- a/word-aligner/support/generate_word_pair_features.pl +++ b/word-aligner/support/generate_word_pair_features.pl @@ -10,9 +10,7 @@ my %fclass = (); load_classes($class_e, \%eclass); load_classes($class_f, \%fclass); -our @IDENT_BINS = qw (Ident0 Ident1 Ident2 Ident3 Ident4 Ident5 Ident6 Ident7 Ident8_9 Ident8_9 Ident10_11 Ident10_11 Ident12_14 Ident12_14 Ident12_14); -die unless scalar @IDENT_BINS == 15; -our $MAX_IDENT_BIN = 'IdentGT' . scalar @IDENT_BINS; +our @IDENT_BINS = qw (Ident0 Ident1 Ident2 Ident3 Ident4 Ident5 Ident6 Ident7 Ident8 Ident9); my $MIN_MAGNITUDE = 0.001; # minimum value of a feature @@ -203,8 +201,8 @@ for my $f (sort keys %fdict) { } } } - my $f_num = ($of =~ /^-?\d[0-9\.\,]+%?$/ && (length($of) > 3)); - my $e_num = ($oe =~ /^-?\d[0-9\.\,]+%?$/ && (length($oe) > 3)); + my $f_num = ($of =~ /^-?\d[0-9\.\,]+%?$/ && (length($of) > 2)); + my $e_num = ($oe =~ /^-?\d[0-9\.\,]+%?$/ && (length($oe) > 2)); my $both_non_numeric = (!$e_num && !$f_num); unless ($total_eandf > 20) { @@ -425,7 +423,8 @@ sub dlenbin { sub identbin { my $x = shift; if ($x == 0) { die; } - if ($x > scalar @IDENT_BINS) { return $MAX_IDENT_BIN; } + $x = int(log($x + 1) / log(1.3)); + if ($x >= scalar @IDENT_BINS) { return $IDENT_BINS[-1]; } return $IDENT_BINS[$x]; } |