search

centre-for-humanities-computing / odyCy

A general-purpose NLP pipeline for Ancient Greek
https://centre-for-humanities-computing.github.io/odyCy/
MIT License
17 stars 2 forks source link

frequency_lemmatizer priority of operations #23

Closed jankounchained closed 1 year ago

jankounchained commented 1 year ago

what order has the best performance?

wanbd runs:

trf_freq_lemmatizer lookup priority. starts above 0.9, manages to get to 0.939

runs 35/36 lemma priority. starts lower, doesn't seem to be able to make it.

can be closed very soon, we basically know that lookup is better. just to mention it somewhere

jankounchained commented 1 year ago

going forward with lookup as the first operation