Extraction of Vietnamese collocation from text corpora

• Main Author: Đỗ, Thị Ngọc Quỳnh
• Format: Theses and Dissertations
• Language: other
• Published: Đại học Quốc gia Hà Nội 2016
• Subjects: Xử lý ngôn ngữ; Xử lý dữ liệu; Ngôn ngữ tự nhiên; Trí tuệ nhân tạo
• Online Access: http://repository.vnu.edu.vn/handle/VNU_123/8281
• Institution: Vietnam National University, Hanoi

Collocations have wide application in the fields of languages, compiled a dictionary as well as the problem of natural language processing. Therefore, the extraction of collocations in each language is really necessary, to improve the accuracy and the nature of the application of natural language processing, as well as help to learn a new language easier. However, in Vietnam, the study of collocation is quite a new field. This paper focused on researching some method of extracting collocations methods to find efficient model for the Vietnamese collocations extraction. The mentioned methods were based on some classic statistical methods commonly used such as frequency, t-test, chi-square, mutual information... We also suggested some general method using linguistic measure to increase the accuracy of the process of extraction. Input data included the data has been through a POS-tagging and data has been parsed. By running the program with different methods and combination of multiple methods together, comparing the accuracy of the method, we draw out the efficient method of extracting of Vietnamese Collocation from Text Corpora.