Cosine similarity calculator

cosine-sim.js

var _ = require('underscore');

var freqVector = function(word, letters) {
  var freq = _.groupBy(word.split(''), function(l) {return l;});
  return _.map(letters, function(l) {
    return freq[l] ? freq[l].length : 0;
  }); 
};

var dot = function(v1, v2) {
  return _.reduce(_.zip(v1, v2), function(acc, els) {
    return acc + els[0] * els[1];
  }, 0); 
};

var mag = function(v) {
  return Math.sqrt(_.reduce(v, function(acc, el) {
    return acc + el * el; 
  }, 0));
};

var f = function(word1, word2) {
  var letters = _.union(word1.split(''), word2.split(''));
  var v1 = freqVector(word1, letters);
  var v2 = freqVector(word2, letters);
  return dot(v1, v2) / (mag(v1) * mag(v2));
};

var args = process.argv;
process.stdout.write(f(args[2], args[3]) + '\n');

// usage:
// $ node cosine-sim.js apple application
// => 0.6416889479197478