Skip to contents

Select best matches per query record

Source: R/select.R
np_select.Rd

Reduces scored, vetoed candidate pairs to the best reference entity per query record, and — because the useful MAYBE/NO review set is wider than a single best guess — also surfaces the best match under two loosened views:

Usage

np_select(pairs, config = np_config(), include_vetoed = FALSE)

Arguments

pairs

Scored + vetoed np_pairs (see np_score(), np_veto()).

include_vetoed

If TRUE, vetoed pairs remain eligible (recorded, not dropped). Default FALSE.

Value

A one-row-per-query data frame of class np_selection with the chosen overall_* match and the alternative name_* / addr_* matches.

Details

  • overall best by combined score (the primary decision),

  • name best by name similarity alone (address ignored beyond blocking),

  • addr best by address similarity alone (name ignored).

When the three views disagree, that disagreement is itself signal for review. Vetoed pairs are excluded from every "best" pick by default.