for each pair of languages should have a single folder from-to (no need for to-from because the cognates file is the same). convention: only have the folder for the alphabetically ordered pair of words (e.g. en-nl, and not nl-en)
- cognates.txt -- format: word_from <space> word_to
- blacklist-<expert-id>.txt -- format: word_from <space> word_to
- rules.txt -- format: pattern_from <space> patter_to
- algo-params.cfg -- simple pairs of key/value (e.g. edit_distance=0.2 [1])
optional:
- experts.txt -- details about the <expert-id>
- params.txt -- params that were used by the candidates generation algorithm when generating cognates
[1] example of config file: https://github.com/zeeguu-ecosystem/Zeeguu-Core/blob/master/default_core.cfg
for each pair of languages should have a single folder from-to (no need for to-from because the cognates file is the same). convention: only have the folder for the alphabetically ordered pair of words (e.g. en-nl, and not nl-en)