Automatic Cognate Detection (acd)

This module provides functions that can be used to evaluate how well algorithms perform in the task of automatic cognate detection.

Functions

bcubes(wordlist[, gold, test, modify_ref, …])

Compute B-Cubed scores for test and reference datasets.

partial_bcubes(wordlist, gold, test[, pprint])

Compute B-Cubed scores for test and reference datasets for partial cognate detection.

pairs(lex[, gold, test, modify_ref, pprint, …])

Compute pair scores for the evaluation of cognate detection algorithms.

diff(wordlist[, gold, test, modify_ref, …])

Write differences in classifications on an item-basis to file.

npoint_ap(scores, cognates[, reverse])

Calculate the n-point average precision.

random_cognates(wordlist[, ref, bias])

Populate a wordlist with random cognates for each entry.

extreme_cognates(wordlist[, ref, bias])

Return extreme cognates, either lump all words together or split them.