cosineSimilarity.rb

require 'benchmark'


d1 = [1,1,1,1,0,0,0,0,0]
d2 = [1,0,1,0,1,1,0,0,0]
d3 = [0,0,0,0,1,0,1,1,1]



def calc_sim vector1, vector2
  a = dot_product vector1, vector2
  b = (normalize vector1) * (normalize vector2)
  a/b
end

def dot_product vector1, vector2
  sum = 0.0
  vector1.each_with_index do | item,  index|
    sum += item * vector2[index]
  end
  sum
end

def normalize vector 
  sum = 0.0
  vector.each do | entry| 
    sum += entry**2
  end
  Math.sqrt sum
end

puts "d1= #{d1.inspect}"
puts "d2= #{d2.inspect}"
puts "d3= #{d3.inspect}"

puts "d1 to d2 #{calc_sim d1,d2}"
puts "d1 to d3 #{calc_sim d1,d3}"
puts "d2 to d3 #{calc_sim d2,d3}"
puts "d1 to d1 #{calc_sim d1,d1}"


puts "doing bigass vector"

bigass1 = []
bigass2 = []

100000.times do |x|
  bigass1 << rand(100) % 2
  bigass2 << rand(100) % 2
end

puts "two #{bigass1.count} entry vectors created. comparing"
result1 = ''
result2 = ''
Benchmark.bmbm do |x|
      x.report("bigass1 with bigass2") { result1 = calc_sim bigass1, bigass2 }
      x.report("bigass2 with bigass1") { result2 = calc_sim bigass2, bigass1}
end

puts "done. ba1 to ba2 #{result1} ba2 to ba1 #{result2}"

# Output
# ruby cosine_similarity.rb 
# d1= [1, 1, 1, 1, 0, 0, 0, 0, 0]
# d2= [1, 0, 1, 0, 1, 1, 0, 0, 0]
# d3= [0, 0, 0, 0, 1, 0, 1, 1, 1]
# d1 to d2 0.5
# d1 to d3 0.0
# d2 to d3 0.25
# d1 to d1 1.0
# doing bigass vector
# two 100000 entry vectors created. comparing
# Rehearsal --------------------------------------------------------
# bigass1 with bigass2   0.270000   0.000000   0.270000 (  0.271032)
# bigass2 with bigass1   0.270000   0.000000   0.270000 (  0.273375)
# ----------------------------------------------- total: 0.540000sec
# 
#                            user     system      total        real
# bigass1 with bigass2   0.270000   0.000000   0.270000 (  0.268135)
# bigass2 with bigass1   0.270000   0.010000   0.280000 (  0.268873)
# done. ba1 to ba2 0.500861220264164 ba2 to ba1 0.500861220264164