Mirandese and Portuguese

Resources for machine translation between Mirandese and Portuguese

Developed Resources

• Corpus Repo
• Mirandese and Portuguese/Evaluation
• Mirandese and Portuguese/Lexical selection
• Mirandese and Portuguese/Contrastive Grammar
• Mirandese and Portuguese/Structural transfer

External Resources

• Mirandese and Portuguese Repo
• Portuguese Transducer
• Mirandese Transducer

Additions

• 5 mwl paradigms
• 3 mwl disambiguation rules
• 3 por disambiguation rules

Final Evaluation

• Monolingual Transducer:
  • Updated precision & recall:
    • Precision: 98.26590%
    • Recall: 57.23906%
  • Coverage over large corpus: ~0.51365440837854797405
  • Number of words in large corpus: 3066116
  • Number of stems in the transducer: 426

• mwl-por:
  • WER and PER over large corpus:
    • WER: 37.82 %
    • PER: 34.24 %
  • Proportion of stems translated correctly in the longer corpus: 315
  • Trimmed coverage over longer and large corpora:
    • Longer:
    • Large: 1574090
  • Number of tokens in longer and large corpora:
    • Longer: 608
    • Large: 3078793