Sentiment analysis based on NLP and deep learning

• Main Author: Lim, Zion Ziheng
• Other Authors: Mao Kezhi
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/167544
• Institution: Nanyang Technological University

Sentiment analysis is a subfield of natural language processing that extracts and identifies sentiments from a string of text. They can be carried out by deep learning models such as RNN, CNN, LSTM, Bi-LSTM and transformer-based models such as BERT, DistilBERT, RoBERTa, XLNET and GPT. This project reviews recent advances in deep learning models for sentiment analysis on datasets that are publicly available. The datasets chosen are from twitter, IMDB, SST2, Yelp and Amazon. We also highlight some of the factors that could affect the performance of the deep learning models such as text representation techniques and hyperparameters. The text representation techniques reviewed are BOW, Word2Vec, GloVe and FastText. Hyperparameters are fine-tuned on the transformer models and their effects can be studied from the results obtained.