sim-hash.rb

  def ngrams(n)
    tokenize.each_with_index.map {|_, i| toks.slice(i, n).join ' ' }
  end

  def sim_hash!
    gramz = Set.new(ngrams(9))

    ordered = gramz.map {|g| g.hash }.sort

    return if ordered.length < 10

    self.sim_hash = [0..10].reduce(0) { |fingerprint, _| fingerprint ^= ordered.shift }
  end