以生成式AI分析金融社群文章情緒對於股市之影響

本研究旨在探討生成式人工智能（AI）分析金融社群文章情緒對於股市的影響，特別聚焦於2021年新冠疫情期間台灣股市的情況。首先利用深度學習和自然語言處理技術，分析了金融社群文章中的直接與間接情緒表達，並探討這些情緒表達對股市的潛在影響。研究結果顯示，不同類型的情緒（如樂觀、悲觀、中立）會顯著影響投資者的決策和市場走勢，尤其在社群媒體與傳統財經新聞相互作用的情況下，這些情緒的影響更加明顯。通過對PTT股票板的文章進行分析，發現社群媒體上的情緒變化與股市指數波動之間存在顯著相關性。具體來說，股市的上漲與正面情緒強烈相關，而當市場出現下跌趨勢時，負面情緒顯著增加。這表明，股市的波動不僅反映了經濟基本面的變化，也可能是社群媒體情緒變化的一個反應器。進一步利用機器學習模型量化情緒分數，並對情緒與股市之間的動態關係進行時序分析。結果顯示，在特定新聞事件或經濟數據公布後，情緒分數與股市指數之間的關聯性會在短時間內顯著增強，這表明社群媒體上的情緒反應可以作為預測股市短期波動的潛在指標。研究發現，即使股市的即時反應可能不顯著，情緒分數對股市影響的異質性，情緒分數的變化在一定時間後仍可能對股市產生影響，在分析社群媒體情緒與股市之間的關聯時，考慮情緒反應的時間是非常必要的。本研究的發現突顯了社群媒體情緒分析在金融市場監測和預測中的應用潛力。生成式AI技術不僅能夠即時捕捉市場情緒變化，還能深化我們對市場動態的理解。在結論中，探討了生成式AI在未來金融市場中的潛在應用，如自動化投資策略的開發和風險評估模型的建立。生成式AI的進步不僅為金融市場分析提供了新的工具，也為金融科技的未來發展開闢了新的道路。未來研究可以進一步探討如何整合更多類型的數據和高級分析技術，提高情緒分析對股市預測的準確性和實用性。考慮社群媒體的全球性和即時性，在全球金融市場動態中的作用亦值得進一步研究。為投資者和金融機構提供更精確和有價值的市場洞察。

關鍵字

生成式AI ；情緒分析；金融社群；股市

並列摘要

This study aims to explore the impact of generative artificial intelligence (AI) analysis of sentiment in financial community articles on the stock market, with a particular focus on the Taiwanese stock market during the COVID-19 pandemic in 2021. By employing deep learning and natural language processing techniques, the research analyzes both direct and indirect expressions of sentiment found in financial community articles and examines their potential influence on the stock market. The results show that different types of sentiments, such as optimism, pessimism, and neutrality, significantly affect investors' decisions and market trends, especially when social media interacts with traditional financial news. Through the analysis of articles on the PTT stock board, a significant correlation was found between changes in sentiment on social media and fluctuations in the stock market index. Specifically, market rises were strongly associated with positive sentiment, while negative sentiment increased significantly during market downturns. This indicates that stock market volatility reflects not only changes in economic fundamentals but could also be a response to shifts in social media sentiment. By further utilizing machine learning models to quantify sentiment scores and conducting time-series analysis on the dynamic relationship between sentiment and the stock market, the results reveal that the correlation between sentiment scores and stock market indices significantly strengthens shortly after specific news events or economic data releases. This suggests that social media sentiment responses can serve as potential indicators for predicting short-term market fluctuations. The findings indicate that even if the immediate reaction of the stock market may not be significant, the heterogeneity of the impact of sentiment scores on the stock market suggests that changes in sentiment scores can still affect the market over time. It is crucial to consider the timing of sentiment reactions when analyzing the relationship between social media sentiment and the stock market. This study highlights the potential application of social media sentiment analysis in monitoring and predicting financial markets. Generative AI technology not only captures market sentiment changes in real-time but also deepens our understanding of market dynamics. In conclusion, the paper discusses the potential applications of generative AI in future financial markets, such as the development of automated investment strategies and the establishment of risk assessment models. The advancement of generative AI not only provides new tools for financial market analysis but also opens new pathways for the future development of financial technology. Future research could further explore how to integrate more types of data and advanced analytical techniques to enhance the accuracy and practicality of sentiment analysis for stock market predictions. Considering the global and real-time nature of social media, its role in global financial market dynamics also warrants further investigation. These studies will help us better understand and respond to market dynamics, providing more precise and valuable market insights for investors and financial institutions.