Machine learning for NTU canteen review analysis and recommendation

• Main Author: Nguyen, Duy Khanh
• Other Authors: Hui Siu Cheung
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/156613
• Institution: Nanyang Technological University

NER, also known as entity identification, chunking, and extraction, is a sub-task of information extraction that aims to discover and categorize named entities referenced in unstructured text into preset categories such as human names, organizations, and locations. Food Named Entity Recognition is the downstream task of NER, which locates, extracts, and classifies food name entities from a sequence of words. Nowadays, there are many methods to extract food name entities from a sentence, such as Terminologydriven, Ruled-based, Corpus-based, Deep Neural Networks based. The Corpus-based method is currently utilized in the FoodHunter project but there are many limitations in this method as it cannot cover all food name entities in Singapore or may be in the world for future improvements. Therefore, this report experiments with Deep Neural Networks, utilizing 2 pretrained models namely T5 and Bart to handle the Food Named Entity Recognition task.