Structured pointing networks for natural language understanding

• Main Author: Nguyen, Thanh Tung
• Other Authors: Joty Shafiq Rayhan
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence; Engineering::Computer science and engineering::Computing methodologies::Document and text processing
• Online Access: https://hdl.handle.net/10356/152207
• Institution: Nanyang Technological University

Natural Language Understanding (NLU) is a subfield in Natural Language Processing (NLP) that deals with machine comprehension of the structure and meaning of human languages. It requires utilizing the relevant context to get the true meanings of the text and how the components of the text are connected with each other to form the structure. Notably, NLU involves finding such linguistic structures and contextual interpretation at the sentence and the document (discourse) level. With the success of deep learning, computers can understand human languages by calculating the relative importance among the inputs to point to the relevant part in the input and then performing inference on those representations. With this technological advancement, this dissertation tackles the NLU problems in two aspects: building better language tools to analyze natural language in terms of explicit representations of the syntax and discourse structures and developing better less-human-effort language modules for many downstream problems.