Text 0.3.2 indel normalizeddistance

Lodestar.Text 0.3.2. This page is frozen at that release. Read the current documentation for what main says now. A link to a decision or a migration page follows main, and leaves the archive.

Indel.NormalizedDistance

Scales the distance into [0, 1] by dividing it by the sum of the two lengths.

public static double NormalizedDistance(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, and it moves both the distance and the lengths it is divided by.

Returns — double in [0, 1]: 0 when the two are equal, 1 when they share nothing at all. Two empty inputs give 0.

Example — five edits over thirteen characters of input.

using Lodestar.Text.Distances;

double d = Indel.NormalizedDistance("kitten", "sitting");   // => 0.3846…

Remarks — the divisor here is len(a) + len(b), not the max(len(a), len(b)) that Levenshtein, Osa and DamerauLevenshtein all use. That is not an inconsistency to work around; it is what keeps the result in [0, 1] for a measure whose raw distance can reach the sum of both lengths rather than the longer of them.

The trap follows from that: a threshold carried over from Levenshtein.NormalizedDistance will be too lenient here, because the same pair of inputs scores lower on this scale. Tune the number against the measure you are actually calling.

Applies to — net10.0, netstandard2.0.

See also — Indel.Distance, Indel.NormalizedSimilarity, Levenshtein.NormalizedDistance, the Python equivalence table.