What's most-frequent-k-characters? It's a metric that measures the similarity between words. (See this Wikipedia article. [yea, Wikipedia, I know])

The authors claim that it's a fast and accurate measure. Unfortunately, I hadn't realized that the paper this algorithm is based on is not peer reviewed. Hence, I'm questioning how useful this is.

I think it might actually be a psuedo metric when defined properly.

Add most-freq-k-chars to your package.json.

 "dependencies": {

    //...

    "most-freq-k-chars": "git://github.com/frankandrobot/most-freq-k-chars.git#master"
  }

Then just do

var freq = require('most-freq-k-chars'),

    rslt = freq.mostFreq2Similarity(word1, word2), //OR
           freq.mostFreq3Similarity(word1, word2, //OR
           freq.mostFreqKSimilarity(word1, word2, K)

As per the authors, #mostFreq2Similarity (K = 2) is the fastest. #mostFreq3Similarity (K = 3) is slightly slower but more accurate. K > 3 is slower and actually less accurate. (So you'll probably want to stick to #mostFreq2Similarity and #mostFrew3Similarity).

I modified the algorithm to match the pseudo-code of the research paper:

With k = 1, #mostFreqKSimilarity

• first it finds the most frequently-occurring letter in both words
• returns the sum of frequencies if the letter occurs in both words; 0 otherwise
• Ex1: mostFreqKSimilarity("aaa", "bbb", 1) = 0
• Ex2: mostFreqKSimilarity("aaBBBB", "BB", 1) = 6

With k = 2, #mostFreqKSimilarity

• first...
  • it finds the most frequently-occurring letter in the 1st word,
  • then it finds the next most frequently-occurring letter in the 1st word
  • it does the same for the 2nd word
• then....
  • returns the sum of the frequencies of the letters that occur in both words; 0 otherwise
• Ex1: mostFreqKSimilarity("aaabb", "xyz", 2) = 0
• Ex2: mostFreqKSimilarity("aaBBBBcc", "BBc", 2) = 9

With k = 3, #mostFreqKSimilarity returns

• first...
  • it finds the most frequently-occuring letter in the 1st word,
  • then it finds the next most frequently-occurring letter in the 1st word
  • then it finds the next most frequently-occurring letter in the 1st word (hence, the k=3)
  • it does the same for the 2nd word
• then....
  • returns the sum of the frequencies of the letters that occur in both words; 0 otherwise
• Ex1: mostFreqKSimilarity("aaabb", "xyz", 3) = 0
• Ex2: mostFreqKSimilarity("aaBBBBcc", "BBc", 3) = 9
• Ex2: mostFreqKSimilarity("aaBBBBcccd", "aBBc", 3) = 6 + 4 + 3

Checkout the tests for more examples.

npm install //you may need to do sudo npm install

jasmin-node tests

• Sadi Evren Seker, Oguz Altun, Uğur Ayan and Cihan MertRada Mihalcea and Paul Tarau, [A Novel String Distance Function based on Most Frequent K Characters] (http://www.researchgate.net/publication/259646670_A_Novel_String_Distance_Function_based_on_Most_Frequent_K_Characters)