Sentiment Analysis of Song Dynasty Classical Poetry Using Fine-Tuned Large Language Models: A Study with LLMs

作者：Baha Ihnaini, Weiyi Sun, Yingchao Cai, Zhijun Xu, Rashid Sangi · 年份：2024 · DOI：10.1109/icaibd62003.2024.10604440 · 被引用次数：9 · 研究领域：Sentiment Analysis and Opinion Mining、Computational and Text Analysis Methods

This study explores the application of advanced large language models (LLMs) in conducting sentiment analysis on classical Chinese literature, focusing on Song Dynasty poetry (Song Ci). The complex linguistic structures and unique emotional expressions inherent in Song Ci pose significant challenges for traditional sentiment analysis methods. Utilizing the sophisticated capabilities of these models, our research applies fine-tuning techniques to navigate and interpret the nuanced language and emotional content of Song Ci more accurately. We evaluate the performance of fine-tuned LLaMA 2 and Qwen models in detecting subtle emotional shifts within these poems. The fine-tuning process incorporates both supervised techniques and reinforcement learning from human feedback, specifically designed to align the models with the historical and cultural context of Song Ci. Notably, the ChatGLM-6B(8-bit) model achieved the best F1 Score of 0.840, demonstrating its exceptional ability to merge ancient literary analysis with modern computational technology, thereby broadening our understanding of the emotional spectrum of classical Chinese poetry.