Vietnamese Sentiment Analysis: An Overview and Comparative Study of Fine-tuning Pretrained Language Models

Sentiment Analysis (SA) is one of the most active research areas in Natural Language Processing (NLP) field due to its potential for business and society. With development language representation models, numerous methods have shown promising efficiency fine-tuning pre-trained models NLP downstream tasks. For Vietnamese, many available were also released, including monolingual multilingual models. Unfortunately, all these trained on different architectures, data, pre-processing steps; consequently, can be expected yield effectiveness. In addition, there no study focusing evaluating performance same datasets SA task up now. This article presents a approach investigate Vietnamese task. The experimental results show superior PhoBERT model ViT5 comparison with previous studies provide new state-of-the-art performances five benchmark datasets. To best our knowledge, first attempt Transformer-based domains sizes