String Similarity

Score any two strings with five classic similarity metrics — Levenshtein, Damerau-Levenshtein, Jaro, Jaro-Winkler, and the Dice bigram coefficient — or use the find-closest mode to rank a query against a candidate list. Two of the metrics are distances (smaller is closer), three are similarities from 0 to 1 (larger is closer). Useful for did-you-mean suggestions, fuzzy dedup, OCR cleanup, and aligning column headers between two CSV files. Everything runs in your browser; nothing is uploaded.

— case-sensitive, character-level
Metrics —
Examples — click any chip to load it
How the metrics differ & when to use each
  • Levenshtein distance counts the minimum number of single-character edits (insert, delete, replace) needed to turn one string into the other. Symmetric, integer, and the workhorse for spelling correction. kitten → sitting is 3.
  • Damerau-Levenshtein distance is Levenshtein plus a single-edit bonus for swapping two adjacent characters, so teh ↔ the counts as 1 instead of 2. Catches the very common transposition typo.
  • Jaro similarity counts matching characters within a sliding match window, then penalises transpositions. Returns 0–1. The canonical example is MARTHA ↔ MARHTA ≈ 0.944. Strong on short strings and names.
  • Jaro-Winkler similarity adds a prefix-length bonus to Jaro, rewarding strings that share an early substring. Best for personal-name and product-name matching, less so for general text.
  • Dice coefficient on character bigrams measures the overlap of two-letter chunks. Fast, well-suited to medium-length strings, and very useful for fuzzy dedup of customer/contact lists or log lines.
  • Distances are sorted ascending (smaller is closer); similarities descending (larger is closer). In find-closest mode the threshold filters out candidates that are obviously too far — set it to a sensible lower bound (similarity) or upper bound (distance) for your use case.
  • All five are case-sensitive by default. Normalise the input (lowercase, trim) before scoring if you want a case-insensitive comparison.
  • Nothing is uploaded. Close the tab and the data is gone.