Trigram similarity index for words.

trigram.rb

require 'minitest/autorun'

class String
  def match_index_to(other)
    this = self.to_trigrams.first
    other = other.to_trigrams.first

    2.0 * ((this & other).size).to_f / (this.size + other.size).to_f
  end

  def to_trigrams
    target = self.downcase.clean_up
    target.chars.each_slice(3).inject([]) { |accum, el| accum << el }
  end

  def clean_up
    self.delete("()[]{}")
  end
end

class TrigramTest < MiniTest::Unit::TestCase
  def setup
    @car_trigrams = [["c", "a", "r"], ["r", "o"]]
    @planet_trigrams = [["p", "l", "a"], ["n", "e", "t"], ["a"]]
  end

  def test_word_to_trigrams
    assert_equal @car_trigrams, "carro".to_trigrams
    assert_equal @planet_trigrams, "planeta".to_trigrams
  end

  def test_remove_garbage
    assert_equal @car_trigrams, "(carro)".to_trigrams
    assert_equal @car_trigrams, "[carro]".to_trigrams
    assert_equal @car_trigrams, "{carro}".to_trigrams
  end

  def test_downcase_string
    assert_equal @car_trigrams, "CARRO".to_trigrams
  end

  def test_match_index
    assert_in_delta 0.66, "carro".match_index_to("barro"), 0.01
    assert_in_delta 1.0, "alexandre".match_index_to("aleksander"), 0.01
    assert_in_delta 0.66, "arthur".match_index_to("artur"), 0.01
    assert_in_delta 0.16, "otorrinolaringologista".match_index_to("barro"), 0.01
    assert_in_delta 0.75, "porra".match_index_to("borra"), 0.01
    assert_in_delta 0.75, "caralho".match_index_to("kralho"), 0.01
  end
end