diff --git a/String/DiceCoefficient.js b/String/DiceCoefficient.js new file mode 100644 index 0000000000..b2f3a34be9 --- /dev/null +++ b/String/DiceCoefficient.js @@ -0,0 +1,51 @@ +/* The Sørensen–Dice coefficient is a statistic used to gauge the similarity of two samples. + * Applied to strings, it can give you a value between 0 and 1 (included) which tells you how similar they are. + * Dice coefficient is calculated by comparing the bigrams of both stings, + * a bigram is a substring of the string of length 2. + * read more: https://en.wikipedia.org/wiki/S%C3%B8rensen%E2%80%93Dice_coefficient + */ + +// Time complexity: O(m + n), m and n being the sizes of string A and string B + +// Find the bistrings of a string and return a hashmap (key => bistring, value => count) +function mapBigrams (string) { + const bigrams = new Map() + for (let i = 0; i < string.length - 1; i++) { + const bigram = string.substring(i, i + 2) + const count = bigrams.get(bigram) + bigrams.set(bigram, (count || 0) + 1) + } + return bigrams +} + +// Calculate the number of common bigrams between a map of bigrams and a string + +function countCommonBigrams (bigrams, string) { + let count = 0 + for (let i = 0; i < string.length - 1; i++) { + const bigram = string.substring(i, i + 2) + if (bigrams.has(bigram)) count++ + } + return count +} + +// Calculate Dice coeff of 2 strings +function diceCoefficient (stringA, stringB) { + if (stringA === stringB) return 1 + else if (stringA.length < 2 || stringB.length < 2) return 0 + + const bigramsA = mapBigrams(stringA) + + const lengthA = stringA.length - 1 + const lengthB = stringB.length - 1 + + let dice = (2 * countCommonBigrams(bigramsA, stringB)) / (lengthA + lengthB) + + // cut 0.xxxxxx to 0.xx for simplicity + dice = Math.floor(dice * 100) / 100 + + console.log('Dice coefficient of', stringA, 'and', stringB, 'is', dice) + + return dice +} +export { diceCoefficient } diff --git a/String/test/DiceCoefficient.test.js b/String/test/DiceCoefficient.test.js new file mode 100644 index 0000000000..f61f5848ba --- /dev/null +++ b/String/test/DiceCoefficient.test.js @@ -0,0 +1,21 @@ +import { diceCoefficient } from '../DiceCoefficient' + +describe('diceCoefficient', () => { + it('should calculate edit distance between two strings', () => { + // equal strings return 1 (max possible value) + expect(diceCoefficient('abc', 'abc')).toBe(1) + expect(diceCoefficient('', '')).toBe(1) + + // string length needs to be atleast 2 (unless equal) + expect(diceCoefficient('a', '')).toBe(0) + expect(diceCoefficient('', 'a')).toBe(0) + + expect(diceCoefficient('skate', 'ate')).toBe(0.66) + + expect(diceCoefficient('money', 'honey')).toBe(0.75) + + expect(diceCoefficient('love', 'hate')).toBe(0) + + expect(diceCoefficient('skilled', 'killed')).toBe(0.9) + }) +})