From 11f476baf7d855198413b091cae775bde4ea41ed Mon Sep 17 00:00:00 2001
From: "olivia.buzek" <olivia.buzek@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Thu, 8 Jul 2010 21:59:50 +0000
Subject: Adding backoff grammar and BackoffRule feature.

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@191 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 gi/pipeline/evaluation-pipeline.pl | 2 ++
 gi/pipeline/local-gi-pipeline.pl   | 8 ++++----
 2 files changed, 6 insertions(+), 4 deletions(-)

(limited to 'gi/pipeline')

diff --git a/gi/pipeline/evaluation-pipeline.pl b/gi/pipeline/evaluation-pipeline.pl
index 619e2699..e0d13340 100755
--- a/gi/pipeline/evaluation-pipeline.pl
+++ b/gi/pipeline/evaluation-pipeline.pl
@@ -14,6 +14,7 @@ my $feat_map = {
   "XFeatures" => [ "XFE","XEF","LabelledEF","LabelledFE"] ,
   "LabelledRuleConditionals" => [ "LabelledFE","LabelledEF" ] ,
   "LexProb" => [ "LexE2F", "LexF2E" ] ,
+  "BackoffRule" => [ "BackoffRule", "BackoffRule" ] ,
 };
 
 my %init_weights = qw(
@@ -27,6 +28,7 @@ my %init_weights = qw(
   LogECount -0.129853
   LogFCount -0.194037
   LogRuleCount 0.256706
+  BackoffRule 0.5
   XFE -0.256706
   XEF -0.256706
   LabelledFE -0.256706
diff --git a/gi/pipeline/local-gi-pipeline.pl b/gi/pipeline/local-gi-pipeline.pl
index c7df2372..99b487db 100755
--- a/gi/pipeline/local-gi-pipeline.pl
+++ b/gi/pipeline/local-gi-pipeline.pl
@@ -69,8 +69,8 @@ safemkdir($OUTPUT) or die "Couldn't create output directory $OUTPUT: $!";
 safemkdir($CONTEXT_DIR) or die "Couldn't create output directory $CONTEXT_DIR: $!";
 safemkdir($CLUSTER_DIR) or die "Couldn't create output directory $CLUSTER_DIR: $!";
 safemkdir($GRAMMAR_DIR) or die "Couldn't create output directory $GRAMMAR_DIR: $!";
-if (-e $TOPICS_CONFIG) {
-  copy($TOPICS_CONFIG, $CLUSTER_DIR) or die "Copy failed: $!";
+if(-e $TOPICS_CONFIG) {
+    copy($TOPICS_CONFIG, $CLUSTER_DIR) or die "Copy failed: $!";
 }
 
 extract_context();
@@ -199,7 +199,7 @@ sub grammar_extract {
   if (-e $OUTGRAMMAR) {
     print STDERR "$OUTGRAMMAR exists, reusing...\n";
   } else {
-    safesystem("$EXTRACTOR -i $LABELED -c $ITEMS_IN_MEMORY -L $BASE_PHRASE_MAX_SIZE | $SORT_KEYS | $REDUCER -p | $GZIP > $OUTGRAMMAR") or die "Couldn't extract grammar";
+    safesystem("$EXTRACTOR -i $LABELED -c $ITEMS_IN_MEMORY -L $BASE_PHRASE_MAX_SIZE -t $NUM_TOPICS -g | $SORT_KEYS | $REDUCER -p | $GZIP > $OUTGRAMMAR") or die "Couldn't extract grammar";
   }
   return $OUTGRAMMAR;
 }
@@ -212,7 +212,7 @@ sub grammar_extract_bidir {
   if (-e $OUTGRAMMAR) {
     print STDERR "$OUTGRAMMAR exists, reusing...\n";
   } else {
-    safesystem("$EXTRACTOR -i $LABELED -c $ITEMS_IN_MEMORY -L $BASE_PHRASE_MAX_SIZE -b | $SORT_KEYS | $REDUCER -p -b | $SORT_KEYS | $REDUCER | $GZIP > $OUTGRAMMAR") or die "Couldn't extract grammar";
+    safesystem("$EXTRACTOR -i $LABELED -c $ITEMS_IN_MEMORY -L $BASE_PHRASE_MAX_SIZE -b -t $NUM_TOPICS -g | $SORT_KEYS | $REDUCER -p -b | $SORT_KEYS | $REDUCER | $GZIP > $OUTGRAMMAR") or die "Couldn't extract grammar";
   }
   return $OUTGRAMMAR;
 }
-- 
cgit v1.2.3