cosine similarity

cosine similarity

def vectorize(msg)
  vector = {}
  words = msg.split(/\s+/)
  words.each do |w|
    h = w.hash
    vector[h] = 1 if vector[h] == nil
    vector[h] += 1
  end
  vector
end

def cosine_similarity(vec1, vec2)
  inner_product = 0.0

  vec1.keys.each do |k|
    if vec2[k] != nil
      inner_product += vec1[k] * vec2[k]
    end
  end

  norm1 = 0.0
  vec1.values.each {|v| norm1 += v ** 2 }
  norm1 = Math.sqrt(norm1)

  norm2 = 0.0
  vec2.values.each {|v| norm2 += v ** 2 }
  norm2 = Math.sqrt(norm2)

  inner_product / (norm1 * norm2)
end