A Review on Advances in Sentiment Analysis: A Deep Learning Approach Using Transformer Based Models
Tejas Vijayrao Kale, Sameer Mendhe
Abstract
A key element of natural language processing is sentiment analysis, which comprises recognizing and understanding opinions and emotions in text. Traditional sentiment categorization methods like machine learning and lexicon-based approaches were made more accurate by deep learning techniques. Transformer-based models that capture long-range relationships through self-attention methods and complicated contextual linkages in text, such as BERT, GPT, RoBERTa, and DistilBERT, have made important contributions to the field. These models perform better than previous methods for multilingual sentiment identification, aspect-based sentiment analysis, and text categorization. However, there are problems like requirement for huge datasets and expensive processing requirements persist. Recent advancements such as lightweight transformer models, multimodal frameworks, and combine efforts of transformer variants like RoBERTa and DistilBERT to improve interpretability address these problems. This study explores the use of RoBERTa-large model for sentiment analysis task using IMDb dataset library containing 50000 movie reviews classified as positive and Negative. RoBERTa-large model achieved accuracy of 96.05% on testing data which shows growing importance of transformer model in NLP applications.