Enhancing The Quality Of Machine Translation System Using Cross Lingual Word Embedding Models

VIETNAM NATIONAL UNIVERSITY, HANOI

UNIVERSITY OF TECHNOLOGY AND ENGINEERING

SOCIALIST REPUBLIC OF VIETNAM

Independence – Freedom – Happiness

INFORMATION ON MASTER’S THESIS

1. Full name: Nguyen Minh Thuan 2. Sex: Male

3. Date of birth: 25/05/1993 4. Place of birth: Hanoi

5. Admission decision number: 393/QĐ-ĐT Dated 21/06/2016

6. Changes in academic process: Change thesis title

7. Official thesis title: Enhancing the quality of Machine Translation System Using Cross￾Lingual Word Embedding Models

8. Major: Computer Science 9. Code: 8480101.01

10. Supervisors: Associate Professor Nguyen Phuong Thai

11. Summary of the findings of the thesis:

My thesis has obtained the following results:

1. Conduct experiments on 3 models of Cross-Lingual Word Embedding to

select the most appropriate for Vietnamese-English and Vietnamese￾Japanese language pairs.

2. Based on the Cross-Lingual Word Embedding models, we propose a model

to improve the quality of the phrase-table in the Phrase-based Machine

Translation System.

3. Based on the Cross-Lingual Word Embedding models, we propose a model

to address the unknown words problem in Neural Machine Translation

system.

4. Results of our experiments shown that the two proposed models enhance the

quality of the machine translation systems.

12. Practical applicability, if any:

The results of our experiments on Vietnamese-English and Vietnamese￾Japanese language pairs indicated that our methods enhance the quality of both

Phrase-based Machine Translation system and Neural Machine Translation system.

Therefore, our methods can be integrated into the state-of-the-art machine

translation system to improve the quality of the translation.