String Similarity
Compare two strings and quantify how alike they are with four standard similarity coefficients. Jaro and Jaro-Winkler measure similarity accounting for character matches within a window and transpositions, with Jaro-Winkler boosting strings that share a common prefix — they are the go-to metric for record linkage and typo-tolerant name matching. Sørensen-Dice and Jaccard compare the character bigram sets of the two strings as set-overlap ratios. All four return 0 (completely different) to 1 (identical). Everything runs locally in your browser.
Similarity scores
Bigrams
Jaro similarity counts characters that match within a sliding window of half the longer string's length, divides the number of out-of-order matches by two to get transpositions t, and returns (m/|a| + m/|b| + (m−t)/m)/3 where m is the match count. Jaro-Winkler adds a prefix bonus: J + ℓ·0.1·(1−J), where ℓ is the length of the common prefix (capped at 4) and the bonus only applies when J > 0.7 (Winkler's boost threshold) — so "MARTHA"/"MARHTA" rises from Jaro 0.9444 to 0.9611. Sørensen-Dice and Jaccard operate on the multisets of character bigrams (overlapping 2-grams): Dice is 2·|A∩B|/(|A|+|B|), Jaccard is |A∩B|/|A∪B|; Dice weights shared bigrams more heavily. Both use multiset (min/max) counts so repeated bigrams are handled correctly. Pairs with the Edit Distance Calculator (Levenshtein/Hamming) and Text Diff tools. Everything runs locally — nothing leaves your browser.