String Similarity

String similarity measures how closely two text strings match each other. It is used in spell checking, DNA analysis, plagiarism detection, and data deduplication.

Levenshtein distance (edit distance) measures the minimum number of single-character edits — insertions, deletions, or substitutions — needed to change one string into another. For example, the distance between "cat" and "bat" is 1.

Jaro-Winkler gives higher scores to strings that share a common prefix. It is widely used for matching names, addresses, and records in databases where typos are common.

Cosine similarity breaks strings into character n-grams, converts them into vectors, and measures the cosine of the angle between the vectors. It works well for longer texts and is commonly used in information retrieval.

Cosine similarity uses n-gram frequency vectors and measures the angle between them, while Jaccard similarity compares sets of n-grams using the ratio of intersection to union. Jaccard ignores duplicates; Cosine does not.

Hamming distance counts the number of positions where corresponding characters differ between two equal-length strings. It only works on strings of the same length and is commonly used in error-correcting codes.

The Sørensen-Dice coefficient (also called Dice's coefficient) measures similarity based on the number of shared bigrams relative to the total bigrams in both strings. It gives more weight to common elements than Jaccard.

For short strings with typos (names, codes), use Levenshtein or Jaro-Winkler. For longer texts, Cosine similarity works best. For quick set-based comparison, use Jaccard. For equal-length strings, Hamming is the simplest.

An n-gram is a contiguous sequence of n characters from a string. For example, the bigrams of "hello" are "he", "el", "ll", "lo". N-grams are the building blocks of many similarity algorithms.

Fuzzy matching finds strings that approximately match a pattern, rather than exactly. It is used in search engines, spell checkers, and data cleaning to handle typos and variations.

No, all calculations are performed locally in your browser. Your text never leaves your device. There is no server-side processing or data storage.

Yes, the algorithms work on any Unicode characters. However, results may vary because different languages have different character structures and word boundaries.