請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/88885
標題: | 應用大型語言模型與提示設計於股市非結構性資料極性分析 Application of Large Language Model and Prompt Design on Stock Market Unstructured Data to Analyze News Polarity |
作者: | 黃冠霖 Kuan-Lin Huang |
指導教授: | 陳永耀 Yung-Yaw Chen |
關鍵字: | 金融情感辭典,成份句法分析,大型語言模型,文本分析,情感分析, Finance sentiment dictionary,Constituency parsing,Large Language Models,Textual analysis,Sentiment analysis, |
出版年 : | 2023 |
學位: | 碩士 |
摘要: | 現代的金融市場中,股票交易已經是普遍的投資方式。股價會隨時間變動而有所漲跌,造成股價變化的原因有許多種,例如政治性因素、經濟性因素或公司性因素等。除了前述提到的因素之外,在網路發達的今日,新聞傳播十分快速,新聞受眾也十分廣泛,從網路取得新聞對一般投資人來說非常的方便。在閱讀新聞後,新聞中所提及的公司相關資訊,會對投資人產生影響,進而做出不同的投資決策,因此新聞對於金融市場波動影響與日俱增。新聞中不僅可以提供數值的資料,更可以提供隱含在文字裡的情緒。而近期在大型語言模型的發展上也取得了大幅的進展,大型語言模型對於文意的理解能力使得對於金融新聞做情感分析的方法可以有新的發展。本研究旨在對網路上的財經新聞進行分析,瞭解使用不同方式對新聞分析的成效。在本文中會使用基於演算法的方式抽取新聞中與特定公司有關的新聞段落對新聞做解析,另外也會使用句法分析及大型語言模型ChatGPT對新聞做分析,並比較三者之間的差異。 In the modern financial market, stock trading has become a common investment method. The stock price will rise or fall over time, and there are many reasons for the stock price change, such as political factors, economic factors or corporate factors. In addition to the factors mentioned above, with the development of the Internet today, news dissemination is very fast, and news audiences are also very wide. Obtaining news from the Internet is very convenient for ordinary investors. After reading the news, the company-related information mentioned in the news will have an impact on investors, and then make different investment decisions. Therefore, the impact of news on financial market fluctuations is increasing day by day. The news can not only provide numerical data, but also provide the emotions implied in the text. Recently, substantial progress has been made in the development of Large Language Models. The ability of Large Language Models to understand the meaning of text enables new developments in the method of sentiment analysis for financial news. This study aims to analyze financial news on the Internet and understand the effectiveness of using different methods to analyze news. In this article, an algorithm-based method will be used to extract news paragraphs related to a specific company in the news to analyze the news. In addition, syntactic analysis and a Large Language Model ChatGPT will be used to analyze the news, and the differences between the three will be compared. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/88885 |
DOI: | 10.6342/NTU202303624 |
全文授權: | 未授權 |
顯示於系所單位: | 電機工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-111-2.pdf 目前未授權公開取用 | 2.15 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。