Word Error Rate

wer.rb

# coding:utf-8
# Word Error Rate (空白が区切り文字前提)

module Eval
  # cost[挿入, 削除, 置換]
  def self.wer( str1, str2, c=[1,1,1] )
    s1 = str1.split(' ')
    s2 = str2.split(' ')

    d = Array.new(s1.size+1) { Array.new(s2.size+1){ 0 } }
    (0..s1.size).each {|n| d[n][0] = n}
    (0..s2.size).each {|n| d[0][n] = n}

    (1..s1.size).each do |i2|
      (1..s2.size).each do |i1|
        cost = s1[i1-1] == s2[i2-1] ? 0 : c[2]

        d[i2][i1] = [d[i2][i1-1] + c[0],   # 挿入
                     d[i2-1][i1] + c[1],   # 削除
                     d[i2-1][i1-1] + cost  # 置換
                    ].min
      end
    end
    return d[s1.size][s2.size] / s1.size.to_f
  end
end