1. Full name:      Nguyen Thi Luong                                            2. Sex: Female

3. Date of birth: 10/10/1983                                                         4. Place of birth: Ha Tinh

5. Admission decision number: No. 633/QĐ-SĐH, dated on 14th March, 2013 by Rector of VNU University of Science, VNU

6. Changes in academic process: Decision No. 636/QĐ-ĐHKHTN dated 14th March, 2014 on the adjustment of Supervisors; Extension education time according to: Decision No. 2786/QĐ-ĐHKHTN dated 31st March, 2016 and  Decision No. 1034/QĐ-ĐHKHTN dated 25th April, 2017; Local return: Decision No.4747/QĐ-ĐHKHTN dated  28th December, 2018  signed by the Rector of VNU University of Science;

7. Official thesis title: Improving the efficiency of Vietnamese syntactic parsing using statistical machine learning approaches.

8. Major: Mathematical Foundation of Informatics               Code: 9460117.02

10. Supervisors: Dr. Le Hong Phuong and Assoc. Prof. Dr. Do Trung Tuan.

11. Summary of the new findings of the thesis

On the linguistic resources: We define a dependent and semantic role schema for Vietnamese. We manually annotate 3,000 sentences with dependency labels and publish the corpus on the universal dependency repository at https://universaldependencies.org/ We annotate a corpus of 5,000 sentences with semantic roles and publish freely for research purpose.

On the methodology: We evaluate the method of self-attentive constituency parsing. We propose to integrate distributed word representations into MSTParser and MaltParser dependency parsers. We offer extensive experimental results on the Bist-parser for Vietnamese dependency parsing. We develop a novel constituent extraction algorithm for improve the quality of semantic role candidates. We integrate distributed word features produced by two recent unsupervised learning models into statistical classifiers and apply integer linear programming inference to improve the accuracy of semantic role labelling for Vietnamese.

12. Practical applicability, if any:

This thesis provides a corpus of thousands of sentences which are annotated with dependency labels and semantic roles for the study of Vietnamese text processing. The thesis develops some fundamental and advanced results in syntactic parsing and semantic role labeling for the Vietnamese language. These results serve as a basis for further study on important practical applications such as automatic question answering, machine translation and others.

13. Further research directions, if any

Enrich the annotated datasets, including dependency corpus and semantic role corpus. Build applications such as machine translation, domain-specific question answering system, information extraction.

14. Thesis-related publications:

Luong Nguyen-Thi, M L. Ha, V H. Nguyen, T M H. Nguyen, and P. Le-Hong, "Building a Treebank for Vietnamese Dependency Parsing", The 10th IEEE RIVF International Conference on Computing and Communication Technologies, Hanoi, Vietnam, IEEE, pp. 147–151, 11/2013.

My, L H., Luong Nguyen-Thi, H N. Viet, T M H. Nguyen, P. Le-Hong, and T H. Phan, "Building a Semantic Role Annotated Corpus for Vietnamese", The 17th National Symposium on Information and Communication Technology, (in Vietnamese), Daklak, Vietnam, pp. 409-414, 11/2014.

Le-Hong, P., T-M-H. Nguyen, Luong Nguyen-Thi, and M-L. Ha, "Fast Dependency Parsing using Distributed Word Representations", Proceedings of PAKDD 2015 Workshops, Trends and Applications in Knowledge Discovery and Data Mining, vol. LNAI 9441, HCM City, Vietnam, Springer, 2015.

Luong Nguyen-Thi, M-L. Ha, P. Le-Hong, and T-M-H. Nguyen, "Using Distributed Word Representations in Graph-Based Dependency Parsing for Vietnamese", The 9th National Conference on Fundamental and Applied Information Technology (FAIR’9), Can Tho, Vietnam, pp. 804–810, 08/2016.

P. Le-Hong, T-H Pham, X-K Pham, T-M-H Nguyen, Luong Nguyen-Thi, M-H Nguyen, "Vietnamese Semantic Role Labeling", VNU Journal of Science: Computer Science and Communication Engineering, Vol. 33, No. 2, pp. 1-21, 2017.

Luong Nguyen-Thi, M-L. Ha, T-M-H. Nguyen and P. Le-Hong, "Using BiLSTM in Dependency Parsing for Vietnamese", Computacion y Sistemas, Vol. 22, No. 3, pp. 853-862, 2018.

Luong Nguyen-Thi, Nguyen Minh Hiep, Dinh Viet Tuan, Phuong Le-Hong “Using Distributed Word Representations in constituency parsing for Vietnamese”,  Proceedings of the conference of information technology, science and technology, Publisher. Science and Technology ISBN: 978-604-67-1191-9, pp 84-90, Nha Trang, 12/2018.

 Luong Nguyen-Thi and Phuong Le-Hong,"An experimental study on constituency parsing for Vietnamese", 16th International Conference of the Pacific Association for Computational Linguistics, Hanoi, Vietnam, 10/2019.