IFuzzyMatcher

Interfaces · AiDotNet.FederatedLearning.PSI

Defines the interface for approximate entity matching in PSI.

For Beginners

In an ideal world, “Patient #12345” at Hospital A is exactly the same string at Hospital B. In practice, one might store “John Smith” and the other “Jon Smith” or “SMITH, JOHN”. Fuzzy matching bridges these gaps by finding IDs that are similar enough to be the same entity.

How It Works

Fuzzy matchers handle the common real-world scenario where entity identifiers aren’t perfectly identical across parties due to typos, formatting differences, transliteration, or data entry errors.

Properties

Property Summary
string StrategyName { get; } Gets the name of this fuzzy matching strategy.

Methods

Method Summary
IReadOnlyList<ValueTuple<int, double>> FindMatches(string id, IReadOnlyList<string> candidates, FuzzyMatchOptions options) Finds all matches for a given identifier from a set of candidates.
bool IsMatch(string id1, string id2, FuzzyMatchOptions options) Determines whether two identifiers are a match according to the configured threshold.
double ComputeSimilarity(string id1, string id2, FuzzyMatchOptions options) Computes the similarity score between two identifiers.
string Normalize(string id, FuzzyMatchOptions options) Normalizes an identifier before comparison by applying case folding, whitespace normalization, and other transformations.