From c97b8a8b58f7385fb48b74e2cf1ea9610cd1202f Mon Sep 17 00:00:00 2001
From: Chris Dyer <redpony@gmail.com>
Date: Mon, 1 Feb 2010 17:38:39 -0500
Subject: word aligner cleanup, new features

---
 word-aligner/support/supplement_weights_file.pl | 73 +++++++++++++++++++++++++
 1 file changed, 73 insertions(+)
 create mode 100755 word-aligner/support/supplement_weights_file.pl

(limited to 'word-aligner/support/supplement_weights_file.pl')

diff --git a/word-aligner/support/supplement_weights_file.pl b/word-aligner/support/supplement_weights_file.pl
new file mode 100755
index 00000000..7f804b90
--- /dev/null
+++ b/word-aligner/support/supplement_weights_file.pl
@@ -0,0 +1,73 @@
+#!/usr/bin/perl -w
+use strict;
+
+my $ADD_FCLASS_JUMP = 1;
+my $ADD_MODEL2_BINARY = 0;
+my $ADD_FC_RELPOS = 1;
+
+my ($f_classes) = @ARGV;
+
+die "Usage: $0 f-classes.file" unless $f_classes && -f $f_classes;
+
+print <<EOT;
+MarkovJump 0
+RelativeSentencePosition 0
+EOT
+
+# !	8
+# "	11
+# 's	18
+
+my %dcats = ();
+$dcats{'BOS'} = 1;
+$dcats{'EOS'} = 1;
+
+open FC, "<$f_classes" or die;
+while(<FC>) {
+  chomp;
+  my ($x, $cat) = split /\s+/;
+  $dcats{$cat} = 1;
+}
+
+my @cats = sort keys %dcats;
+
+my $added = 0;
+for (my $i=0; $i < scalar @cats; $i++) {
+  my $c1 = $cats[$i];
+  for (my $j=0; $j < scalar @cats; $j++) {
+    my $c2 = $cats[$j];
+    print "SP:${c1}_${c2} 0\n";
+    $added++;
+  }
+}
+
+for (my $ss=1; $ss < 100; $ss++) {
+  if ($ADD_FCLASS_JUMP) {
+    for (my $i=0; $i < scalar @cats; $i++) {
+      my $cat = $cats[$i];
+      for (my $j = -$ss; $j <= $ss; $j++) {
+        print "Jump_FL:${ss}_FC:${cat}_J:$j 0\n";
+        $added++;
+      }
+    }
+  }
+  if ($ADD_MODEL2_BINARY) {
+    # M2_FL:8_SI:3_TI:2=1
+    for (my $i = 0; $i < $ss; $i++) {
+      for (my $j = 0; $j < 100; $j++) {
+        print "M2_FL:${ss}_SI:${i}_TI:${j} 0\n";
+        $added++;
+      }
+    }
+  }
+}
+if ($ADD_FC_RELPOS) {
+  #RelPos_FC:11
+  for (my $i=0; $i < scalar @cats; $i++) {
+    my $cat = $cats[$i];
+    print "RelPos_FC:$cat 0\n";
+    $added++;
+  }
+}
+
+print STDERR "Added $added weights\n";
-- 
cgit v1.2.3