LARGE LANGUAGE MODEL AND THEIR IMPACT ON SENTIMENT ANALYSIS

Abstract

The advent of large language models (LLMs) has significantly advanced the field of natural language processing (NLP), particularly in the area of sentiment analysis. This thesis explores the impact of LLMs on sentiment analysis, focusing on their ability to accurately classify and interpret human emotions in textual data. Utilizing LLM model such as BERT, this research examines how these advanced architectures improve sentiment classification in terms of precision, recall, and F1-score, compared to traditional machine learning techniques. Through comprehensive experimentation and analysis, we demonstrate the efficacy of LLMs in sentiment analysis tasks. For instance, using the BERT model on the Twitter US Airline Sentiment dataset, we achieved impressive classification metrics, including a precision of 1.00 for negative sentiment, and high overall scores in micro, macro, and weighted averages. The confusion matrix further illustrates BERT's capability to correctly classify sentiments, with minimal misclassifications across negative, neutral, and positive categories. This study also addresses the challenges associated with deploying LLMs, such as computational demands, model interpretability, and ethical considerations. Additionally, we explore the practical applications of LLMs in various domains, including social media monitoring, customer feedback analysis, and market research. The findings of this thesis underscore the transformative potential of large language models in enhancing sentiment analysis, providing valuable insights for researchers and practitioners aiming to leverage these models for more accurate and nuanced emotion detection.