Meta-evaluation

From Apertium
Revision as of 22:01, 30 May 2019 by Firespeaker (talk | contribs)
Jump to navigation Jump to search

Apertium language modules and translation pairs are subject to the following types of evaluation:


  • Naïve coverage
    • Monolingual naïve coverage
    • Trimmed naïve coverage (i.e., using a trimmed dictionary)
  • Accuracy of tagging
    • Precision/Recall/F-score
  • Accuracy of translation
    • WER/PER/BLEU
  • Clenliness of translation output
    • Testvoc