Text jaro similarity

Development build. This page describes main, not a released package. The latest published Lodestar.Text is 0.4.0 — read its documentation.

Jaro.Similarity

Scores two strings on how many characters they share within a sliding window, and how many of those shared characters arrive in a different order.

public static double Similarity(ReadOnlySpan<char> a, ReadOnlySpan<char> b, TextElement element = TextElement.Utf16Unit)

Parameters — a and b are the two strings to compare. element says what counts as one character; jellyfish works on code points, so pass TextElement.CodePoint to reproduce its numbers on supplementary-plane text.

Returns — double in [0, 1], larger meaning more alike. 1 for equal non-empty inputs — see the trap below for what two empty ones give.

Example — one transposition in a six-letter name barely dents the score.

using Lodestar.Text.Distances;

double s = Jaro.Similarity("MARTHA", "MARHTA");   // => 0.9444…

Remarks — Jaro was designed for matching people's names in record linkage, and that is still what it is best at: short strings, a handful of characters, where a typo or a swap should barely move the score. It is far more forgiving than Levenshtein on exactly those inputs, and far less meaningful on long text, where the matching window grows with the length and almost everything ends up "near" something.

The trap is the empty case, and it is the opposite of what the rest of this page does. Two empty strings score 0, not 1 — Jaro.Similarity("", "") reports no similarity at all. That is jellyfish's convention and this implementation follows it deliberately, but it means an empty field never matches another empty field here, while RatcliffObershelp.Similarity("", "") returns 1. A single input being empty also gives 0.

Applies to — net10.0, netstandard2.0.

See also — Jaro.Distance, JaroWinkler.Similarity, decision 0005, the Python equivalence table.