Sentiment Analysis of Twitter Data Using NLP Models: A Comprehensive Review

Social media platforms, particularly Twitter, have become vital sources for understanding public sentiment due to the rapid, large-scale generation of user opinions. Sentiment analysis of Twitter data has gained significant attention as a method for comprehending public attitudes, emotional responses, and trends which proves valuable in sectors such as marketing, politics, public health, and customer services. In this paper, we present a systematic review of research conducted on sentiment analysis using natural language processing (NLP) models, with a specific focus on Twitter data. We discuss various approaches and methodologies, including machine learning, deep learning, and hybrid models with their advantages, challenges, and performance metrics. The review identifies key NLP models commonly employed, such as transformer-based architectures like BERT, GPT, etc. Additionally, this study assesses the impact of pre-processing techniques, feature extraction methods, and sentiment lexicons on the effectiveness of sentiment analysis. The findings aim to provide researchers and practitioners with a comprehensive overview of current methodologies, insights into emerging trends, and guidance for future developments in the field of sentiment analysis on Twitter data.