Winklestein

Winklestein is a hybrid string similarity algorithm that combines Levenshtein Distance and Jaro–Winkler Similarity to provide accurate, robust, and tunable similarity scoring for short and medium-length text inputs.

It is especially effective for:

• Phone numbers
• Names
• Addresses
• Fuzzy matching
• Record deduplication
• Autocorrect and search suggestions

---

📦 Packages

Package	Release	Preview	Downloads
Xreeple.Winklestein

🚀 Features

• Hybrid scoring algorithm (Levenshtein + Jaro–Winkler)
• Normalized similarity output between 0.0 and 1.0
• Three configurable tolerance modes:
  • Aggressive — More forgiving, marks more strings as similar
  • Normal — Balanced, recommended for most cases
  • Lenient — Strict, lower similarity unless strings are very close
• Zero-dependency, cross‑platform .NET Standard library
• Simple one‑line usage:

  var score = Winklestein.Compare("hello", "h3llo");

---

📦 Installation

dotnet add package Winklestein

---

🧠 Similarity Modes

Winklestein exposes three tunable modes that modify weighting and output sensitivity:

🔴 Aggressive

• More tolerant of differences
• Boosts similar prefixes strongly
• Useful for:
  • OCR corrections
  • Contact search
  • Misspelled user input
• Produces higher similarity scores

Example:

"mehmet" vs "mehmed" → higher score

---

⚪ Normal (default)

• Balanced weighting
• Works for most fuzzy‑match comparisons
• Recommended for general-purpose applications

---

🔵 Lenient

• Stricter and less forgiving
• Levenshtein penalty is increased
• Jaro–Winkler boost is reduced
• Suitable when:
  • High precision is needed
  • Probabilistic matching must be conservative

Example:

"mehmet" vs "mehmed" → lower score

---

🧩 Example Usage

Basic comparison

using Winklestein;

var result = Winklestein.Compare("hello", "h3llo");
Console.WriteLine(result);   // 0.89 (example)

With mode selection

using Winklestein;

var result = Winklestein.Compare("mehmet", "mehmed", SimilarityMode.Aggressive);

All modes

var a = Winklestein.Compare("metro", "metre", SimilarityMode.Aggressive);
var b = Winklestein.Compare("metro", "metre", SimilarityMode.Normal);
var c = Winklestein.Compare("metro", "metre", SimilarityMode.Lenient);

---

🧪 Unit Tests

You can find full tests in the repository, including:

• Basic equality checks
• Mode sensitivity checks
• Distance and Winkler accuracy tests

---

⚡ Benchmarking

Benchmarks use BenchmarkDotNet and include:

• Small strings
• Long strings
• High‑difference strings
• Mode comparisons

Run with:

dotnet run -c Release

---

📜 License

MIT License.

---

💬 Contribution

Contributions, issues, and PRs are welcome!