Fuzzy Wuzzy Algorithm
A string matching algorithm built on top of Levenshtein distance, designed specifically for finding approximate matches in messy real-world data.
Fuzzy Wuzzy is not available in Tilores due to licensing restrictions. You can explore it independently via the thefuzz library on GitHub (formerly FuzzyWuzzy by SeatGeek).
How it works
Fuzzy Wuzzy wraps Levenshtein distance with additional preprocessing and several matching strategies. Rather than comparing strings directly, it normalises them first — lowercasing, stripping punctuation, and trimming whitespace — before computing similarity scores.
It offers four matching modes: a simple ratio (standard Levenshtein similarity), a partial ratio (matches the shorter string against substrings of the longer one), a token sort ratio (sorts tokens alphabetically before comparing, useful for reordered names), and a token set ratio (compares the intersection and remainder of token sets, useful for subset matching). Scores are returned as integers from 0 to 100.
Use cases in entity resolution
Related tools
Don't implement this yourself
Tilores Studio runs multiple matching algorithms — with configurable rules and real-time entity resolution — locally on your machine. Free, no account, no cloud. Load your own data and see it working in minutes.