Atsushi FUJITA, Dr. Eng.

Research Interests

Research Fields:
Computational Linguistics, Natural Language Processing
Research Topics:
Automatic Paraphrasing, Machine/Human Translation, Others
Interests:
Machine Learning, Data Mining, Social Network Analysis, Text Annotation, Language Communication Support, etc.

Projects and Recent Publications/Presentations

Multilingual/Cross-lingual NLP (mainly Machine Translation)

We are conducting research on machine translation and multilingual text processing from various aspects. Mainly with my colleagues at NICT and sometimes in collaboration with researchers outside NICT, including internship students.

  • [Rubino+, 2021] Eval4NLP. Error Identification for Machine Translation with Metric Embedding and Attention. (Best Overall Approach Award)
  • [Dabre & Fujita, 2021] MT Summit. Investigating Softmax Tempering for Training Neural Machine Translation Models.
  • [Marie+, 2021] ACL-IJCNLP. Scientific Credibility of Machine Translation Research: A Meta-Evaluation of 769 Papers. (Outstanding Paper Award)
  • [Rubino+, 2020] MT Journal. Extremely Low-Resource Neural Machine Translation for Asian Languages.
  • [Dabre & Fujita, 2020] WMT. Combining Sequence Distillation and Transfer Learning for Efficient Low-Resource Neural Machine Translation Models.
  • [Marie & Fujita, 2020] TACL. Synthesizing Parallel Data of User-Generated Texts with Zero-Shot Neural Machine Translation.

Dissecting and Modeling Translation Process

A multi-institutional KAKENHI project "Developing a Translation Process Model and Constructing an Integrated Translation Environment through Detailed Descriptions of Translation Norms and Competences" led by Prof. Kyo Kageura.

  • [Yamamoto+, 2021] IATIS. Designing a Metalanguage of Translation Strategies for Translation Training: Demystifying the Art of Translation.
  • [Fujita, 2021] MT Summit. Attainable Text-to-Text Machine Translation vs. Translation: Issues Beyond Linguistic Processing.
  • [Miyata & Fujita, 2021] EACL. Understanding Pre-Editing for Black-Box Neural Machine Translation.