From 2b1d7f881c19c4d4b5afae194e02d3300c7675d0 Mon Sep 17 00:00:00 2001
From: Patrick Simianer <p@simianer.de>
Date: Tue, 5 Jul 2016 11:01:46 +0200
Subject: mv

---
 kbest-bleu-oracles | 51 +++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)
 create mode 100755 kbest-bleu-oracles

(limited to 'kbest-bleu-oracles')

diff --git a/kbest-bleu-oracles b/kbest-bleu-oracles
new file mode 100755
index 0000000..a36c345
--- /dev/null
+++ b/kbest-bleu-oracles
@@ -0,0 +1,51 @@
+#!/usr/bin/env ruby
+
+require 'zipf'
+require 'trollop'
+
+def get_context kbest_lists, references, n
+  a = []
+  kbest_lists.each_index { |i|
+    a << BLEU::get_counts(kbest_lists[i][0].s, references[i], n, 1)
+  }
+  return a
+end
+
+def main
+  conf = Trollop::options do
+    opt :kbest_lists, "kbest lists", :type => :string, :required => true
+    opt :references, "reference", :type => :string, :required => true
+    opt :n, "N for BLEU", :type => :int, :default => 4
+    opt :weight, "how much to weigh single translations", :type => :int, :default => 1
+    opt :debug, "debug mode", :type => :bool, :default => false
+  end
+  debug = conf[:debug]
+  n = conf[:n]
+  kbest_lists = read_kbest_lists conf[:kbest_lists]
+  references = ReadFile.readlines_strip conf[:references]
+  context = get_context kbest_lists, references, n
+  kbest_lists.each_with_index { |kbest,j|
+    scores = []
+    max_score = -1.0/0
+    max_idx = -1
+    kbest.each_index { |i|
+      context_cp = context.dup
+      context_cp[j] = BLEU::get_counts kbest[i].s, references[j], n, conf[:weight]
+      score = BLEU::hbleu_(context_cp, n, debug)
+      scores << score
+      if score > max_score
+        max_score = score
+        max_idx = i
+      end
+      STDERR.write "#{i} #{kbest[i]}\t#{score}\n---\n" if debug
+    }
+    puts "#{references[j]}"
+    puts "BLEU=#{scores[0]} ||| #{kbest[0]}"
+    puts "BLEU=#{max_score} ||| #{kbest[max_idx]}"
+    puts
+    STDERR.write "\n" if debug
+  }
+end
+
+main
+
-- 
cgit v1.2.3