Fuzzy-match two sets of species names

Uses component-based similarity: genus and epithet are matched separately, then combined with weights (genus 0.6, species epithet 0.4; at subspecies rank, genus 0.55, species epithet 0.30, and infraspecific epithet 0.15). Uses base R utils::adist() for Levenshtein distance, with no extra dependencies.

Usage

pr_fuzzy_match(
  names_x,
  names_y,
  threshold = 0.9,
  rank = "species",
  quiet = FALSE,
  identity_x = NULL,
  identity_y = NULL
)

Arguments

names_x

Character vector.

names_y

Character vector.

threshold

Numeric (0–1). Minimum similarity score. Default 0.9.

rank

Character. "species" or "subspecies".

quiet

Logical. Suppress progress messages.

identity_x

Internal authority identity evidence aligned to names_x. Used only by the reconciliation cascade.

identity_y

Internal authority identity evidence aligned to names_y. Used only by the reconciliation cascade.

Value

A tibble with columns: name_x, name_y, score, notes.

Details

Candidate generation starts with names whose genus is within 2 edits. When comparable authority identity is available, same-ID names in other genera can also be considered. The existing pr_suggest_gender() helper identifies its limited -us/-a/-um and -is/-e ending pairs. These pairs are withheld without identity evidence and can match only when provider, snapshot, rank, and accepted ID agree. The adjusted score treats the supported species-epithet ending as equivalent; at subspecies rank, the infraspecific epithet must also be sufficiently similar. This does not establish grammar or identity. Other fuzzy matches retain their spelling-based behavior.

The identity_x and identity_y arguments are internal evidence inputs supplied by the reconciliation cascade. They are not part of the exported user interface.