Refining the state-of-the-art in Machine Translation, optimizing NMT for the JA <-> EN language pair by leveraging personal domain expertise

TL;DR

This paper details the development and optimization of a Neural Machine Translation system for Japanese-English translation using Transformer architecture, focusing on corpus processing, hyperparameter tuning, and evaluation metrics.

Contribution

It introduces a systematic approach to optimize NMT performance for En/Ja by combining technical tuning with linguistic expertise.

Findings

Achieved improved BLEU scores for En/Ja translation

Demonstrated the impact of corpus preprocessing on translation quality

Provided subjective linguistic evaluation alongside standard metrics

Abstract

Documenting the construction of an NMT (Neural Machine Translation) system for En/Ja based on the Transformer architecture leveraging the OpenNMT framework. A systematic exploration of corpora pre-processing, hyperparameter tuning and model architecture is carried out to obtain optimal performance. The system is evaluated using standard auto-evaluation metrics such as BLEU, and my subjective opinion as a Japanese linguist.