#!/bin/sh

hadoop jar /usr/lib/hadoop-0.20-mapreduce/contrib/streaming/hadoop-streaming-2.0.0-mr1-cdh4.1.2.jar \
  -D mapred.reduce.tasks=24 \
  -D mapred.task.timeout=1200000000 \
  -input overlap/newstest2008-grammar+index \
  -output overlap/newstest2008-grammar_word-pair-keys \
  -mapper word_pair_keys.rb  \
  -reducer "/bin/cat" \
  -file word_pair_keys.rb \
  -file util.rb