Lodestar.Text 0.6.0. 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.

The fingerprint of a bag of tokens.

public static ulong Fingerprint(IEnumerable<string> tokens)
public static ulong Fingerprint(IEnumerable<KeyValuePair<string, int>> weighted)

Parameters — tokens are weighted once each, repeats included. weighted carries its own non-negative weight per token, which is the overload to use when a term frequency is already known.

Returns — ulong, a 64-bit fingerprint. An empty input gives zero.

Exceptions — ArgumentNullException when the sequence, or a token in it, is null; ArgumentOutOfRangeException when a weight is negative.

Example — repeating a token moves the fingerprint.

using Lodestar.Text.Similarity;

ulong once = SimHash.Fingerprint(["the", "quick", "brown", "fox"]);
ulong twice = SimHash.Fingerprint(["the", "the", "quick", "brown", "fox"]);

bool moved = once != twice;  // => True

Remarks — each token is hashed with MD5, read as a big-endian integer, and its low 64 bits decide which columns it pushes up and which down; a column that ends positive sets its bit. MD5 is a hash function here and never a signature, and every frozen value depends on it, so a stronger digest would be a different algorithm rather than an improvement.

A weight of zero is not the same as omitting the token. It contributes nothing to any column, which is what omitting it does too — but it still has to be hashed, so passing a long tail of zero-weight tokens costs time and buys nothing.

Applies to — net10.0, netstandard2.0.

See also — SimHash.HammingDistance.