From 6bd7135e6039b0682f49234e42451077413f0bd9 Mon Sep 17 00:00:00 2001
From: Patrick Simianer <p@simianer.de>
Date: Tue, 10 May 2016 10:49:09 +0200
Subject: count clicks and keystrokes, fix bug in rule addition, good params
 and improvement for phrase2 extraction

---
 phrase2_extraction/phrase2_extraction.rb | 38 ++++++++++++++++++++++++++++----
 1 file changed, 34 insertions(+), 4 deletions(-)

(limited to 'phrase2_extraction')

diff --git a/phrase2_extraction/phrase2_extraction.rb b/phrase2_extraction/phrase2_extraction.rb
index 253df1b..547e0be 100755
--- a/phrase2_extraction/phrase2_extraction.rb
+++ b/phrase2_extraction/phrase2_extraction.rb
@@ -5,9 +5,9 @@ require 'zipf'
 module PhrasePhraseExtraction
 
 DEBUG                      = false
-MAX_NT                     = 1    # Chiang: 2
-MAX_SEED_NUM_WORDS         = 10   # Chiang: 10 words
-MAX_SRC_SZ                 = 5    # Chiang: 5 words
+MAX_NT                     = 2    # Chiang: 2
+MAX_SEED_NUM_WORDS         = 4    # Chiang: 10 words, -> phrases!
+MAX_SRC_SZ                 = 10   # Chiang: 5 words, -> words!
 FORBID_SRC_ADJACENT_SRC_NT = true # Chiang:true
 
 class Rule
@@ -51,6 +51,21 @@ class Rule
     return src_len
   end
 
+  def len_src_w
+    src_len = 0
+    @source.each { |i|
+      if i.is_a? String
+        src_len += i.split.size #1
+      else
+        i.each { |j|
+          src_len += source_context[j].split.size
+        }
+      end
+    }
+
+    return src_len
+  end
+
   def len_tgt
     tgt_len = 0
     @target.each { |i|
@@ -64,6 +79,21 @@ class Rule
     return tgt_len
   end
 
+  def len_tgt_w
+    tgt_len = 0
+    @target.each { |i|
+      if i.is_a? String
+        tgt_len += i.split.size
+      else
+        i.each { |j|
+          tgt_len += target_context[j].split.size
+        }
+      end
+    }
+
+    return tgt_len
+  end
+
   def to_s
     source_string = ""
     @source.each { |i|
@@ -625,7 +655,7 @@ end
 
 def PhrasePhraseExtraction.remove_too_long_src_sides rules
   return rules.reject { |r|
-    r.len_src > PhrasePhraseExtraction::MAX_SRC_SZ
+    r.len_src_w > PhrasePhraseExtraction::MAX_SRC_SZ
   }
 end
 
-- 
cgit v1.2.3