DSpace at VNU: Building a semantic role labeling system for Vietnamese sentences

Building a semantic role labeling system for
Vietnamese sentences
Nguyễn Thanh Huy
Trường Đại học Công nghệ
Luận văn Thạc sĩ ngành: Khoa học máy tính; Mã số: 60 48 01
Người hướng dẫn: PhD. Nguyễn Phương Thái
Năm bảo vệ: 2011
Keywords: Nhận dạng ký tự; Tiếng Việt; Tin học
Abtract: Function tags are labels which are defined as labels which contain abstract
syntactic information of constituents such as noun phrases and prepositional phrase in a
sentence. Beside syntactic labels, function tags have significant roles in Nature Language
Processing (NLP). Function tags are the key to answer several questions such as: who,
whom, where, what, which often use in problems as: question answering, information
extraction, and summarization. These questions are very hard to answer if a NLP-system
has only syntactic information. In recent years, function tag labeling has been studied for
languages such as English and Chinese. In Vietnam, Nature Language Processing
researches have recently studied in parsing syntactic problems. In this thesis, I focus on
building a function tag labeling system for Vietnamese. With our experiment, I hope that
our system will enrich data sources for NLP-research in Vietnamese later. The structure
of thesis includes three parts. In the first part, I introduce function tags: their applications
and why they attract researchers to explore the relationship between function tags with
constituents in a sentence. I also mention some modern research in function tag labeling
problem including: related works, features, approaches and methods. Besides, I will
introduce Vietnamese function tags, corpus Viet Treebank, which is used in many NLP
research in Vietnamese.In the second part, I represent function tag labeling approaches in
recent research. Briefly, for each approach I will propose and discuss aspects such as:
approach strategy, features, methods, advantages and disadvantages. For other methods
the reader can refer to the bibliography. In the last part, I will present our research in
function tag labeling for Vietnamese. I will discuss related works such as: processes to
prepare data sources, selection features, proposal model and the effectiveness of our
method.

Content