java-string-similarity

Implements over a dozen classic algorithms like Levenshtein, Jaro-Winkler, and cosine similarity, covering most common string comparison needs without requiring external implementations.

Defines interfaces for normalized vs. metric distances, as detailed in the README's table, helping users select algorithms based on mathematical properties like triangle inequality.

Supports shingle-based methods (e.g., Q-Gram, Cosine) that allow pre-computing string profiles for O(m+n) comparisons, ideal for large datasets as shown in the precomputed cosine example.

Offers weighted Levenshtein for OCR/keyboard correction and optimal string alignment for restricted edit distance, addressing niche use cases directly mentioned in the documentation.