Levenshtein Distance — Compare Two Strings

What is Levenshtein Distance?

Levenshtein distance (also called edit distance) is the smallest number of single-character edits needed to turn one string into another. The allowed edits are insertion, deletion, and substitution. For example, turning kitten into sitting takes three edits (k→s, e→i, insert g), so the distance is 3.

Why people use it

• Spell checking & fuzzy search: find words that are “close enough” to a query.
• Data cleaning: match near-duplicate names, addresses, or product titles.
• Natural language & bioinformatics: compare tokens, lines, or genetic sequences.
• UX matching: tolerate typos in forms and search boxes.

Character vs. Word mode

• Character mode: great for short strings, usernames, variable names, and typos.
• Word mode: splits on whitespace and compares tokens, which is clearer for sentences and paragraphs. It answers “how many word-level changes are needed?”

Similarity score

We show a simple similarity percentage: 1 − (distance / max(length(A), length(B))). In word mode, “length” means word count.

Case, Unicode, and language notes

• Case sensitivity: toggle on/off. For case-insensitive matching, we compare lowercase forms but keep your original text for display.
• Unicode: operates on code points. Some grapheme clusters (e.g., emoji + modifiers) may count as multiple edits.

How it’s computed (light overview)

We use dynamic programming over a (len(A)+1) × (len(B)+1) grid. The bottom-right cell is the distance; backtracking yields the sequence of edits we highlight.

Related metrics

• Damerau–Levenshtein: counts adjacent transpositions as one edit.
• Jaro / Jaro–Winkler: emphasize matching characters and order.
• Cosine / Jaccard on tokens: compare word sets or vectors for document similarity.