Inputs: Two names. We normalize (lowercase, strip spaces/punctuation, basic accent fold).
Score (0–100) combines:
Shared letters (0–40): overlap ratio across unique characters.
Bigrams overlap (0–25): Jaccard index of adjacent-letter pairs.
Name similarity (0–25): normalized Levenshtein similarity.
Tie-break (0–20): stable hash-based component for variety.
Labels help interpret the % (e.g., “Exceptional match” ≥ 90%).
Deterministic: Same names → same score. This is for entertainment only.
Privacy: Drag & drop/file pick is parsed locally in your browser; nothing is uploaded.
Tips & notes
Try nicknames vs full legal names — normalization reduces punctuation/spacing impact.
Very short names have fewer bigrams; the algorithm still scales but expect lower overlap.
The score is not a prediction of real relationships — it’s a consistent, mathy ice-breaker.
Cookies & privacy
Utilities Bunker uses necessary cookies to run the site. Optional cookies help remember preferences,
measure usage, and improve the platform. You can change your choices any time.
Privacy Policy.
