外匯市場是全球流動性最強的金融市場，匯率波動對企業和國家的經濟競爭力有重要影響，不論企業是否進行跨國交易，匯率波動都會影響其經營成本和利潤。隨著網路技術的發展，網路新聞成為影響外匯市場的重要因素，然而人工分析大量新聞的速度常常落後於市場反應，導致決策延誤。因此，本研究運用文字探勘技術分析網路新聞情緒，以預測隔日歐元兌美元的匯率變動，並探討其有效性。
研究期間為2020年1月1日至2023年12月29日的歐元兌美元收盤匯率資料，並從Investing.com蒐集相應期間內的「FOREX」分類之新聞文章，利用Bert-base進行情感分析，將情緒分數納入VARMA和LSTM模型中進行匯率預測，並評估加入新聞情緒前後兩模型的預測效果。研究發現，加入新聞情緒變數後，LSTM模型的預測準確度顯著提高，優於傳統的VARMA模型。因此，新聞情緒對匯率的波動具有有效的預測能力，能夠有效預測隔日歐元匯率。
本研究證實了新聞情緒在預測匯率變動中的重要性，並表明LSTM模型結合情緒分析在匯率預測中的優越性，企業可利用這一方法應對市場變化，減少匯率風險。本研究提供了文字探勘技術在金融市場中的應用範例，對未來的研究具有參考價值。

The foreign exchange market is the most liquid financial market globally, with exchange rate fluctuations significantly impacting economic competitiveness. These fluctuations affect operating costs and profits, regardless of cross-border transactions. Online news has become a crucial factor influencing the market, but manual news analysis often lags behind market reactions, causing delayed decisions. This study uses text mining techniques to analyze online news sentiment to predict next-day EUR/USD exchange rate changes and assess its effectiveness.
The research covers EUR/USD closing exchange rate data from January 1, 2020, to December 29, 2023, and collects news articles categorized under "FOREX" from Investing.com. Sentiment analysis is conducted using Bert-base, with sentiment scores incorporated into VARMA and LSTM models for exchange rate prediction. The study evaluates model performance with and without news sentiment. Results show that incorporating news sentiment significantly enhances the LSTM model's prediction accuracy, outperforming the traditional VARMA model. Thus, news sentiment effectively predicts exchange rate fluctuations and the next day's EUR exchange rate.
This study confirms the importance of news sentiment in predicting exchange rate changes and demonstrates the superiority of combining sentiment analysis with the LSTM model. Businesses can use this method to respond to market changes and reduce exchange rate risks. This study provides a valuable example of applying text mining techniques in the financial market, offering reference value for future research.

論文審定書 i
誌謝 ii
摘要 iii
Abstract iv
目錄 v
圖次 vii
表次 viii
第一章、 緒論 1
第一節、 研究背景 1
第二節、 研究動機 2
第三節、 研究目的 3
第二章、 文獻探討 5
第一節、 文字探勘與情緒分析 5
第二節、 新聞情緒對於市場影響 5
第三節、 BERT 6
第四節、 長短期記憶模型(LSTM) 9
第三章、 研究方法 14
第一節、 研究標的 15
第二節、 解釋變數 15
(一). 通膨差異（INF） 15
(二). 實質即期利率差分（INT） 17
(三). 恐慌指數（Volatility Index，VIX） 17
(四). MSCI Europe Index（MSCI） 17
第三節、 新聞情緒（News_sent） 19
第四節、 實驗模型 23
(一). 傳統時間序列模型－向量自回歸移動平均模型（VARMA） 23
(二). 神經網路模型－長短期記憶模型（LSTM） 26
第五節、 評估預測表現 27
(一). 平均平方誤差（Mean-Square Error , MSE） 28
(二). 均方根誤差（Root Mean Squared Error , RMSE） 28
(三). 平均絕對誤差（Mean Absolute Error , MAE） 28
(四). 平均絕對百分比誤差（Mean Absolute Percentage Error , MAPE） 29
第四章、 實驗分析 30
第一節、 資料敘述統計 30
第二節、 平穩性檢定 35
第三節、 實驗模型分析結果 37
第五章、 結論與建議 42
第一節、 研究結論 42
第二節、 研究限制與建議 43
參考文獻 44

1. Adebiyi, A., Adewumi, A., & Ayo, C. (2014). Stock price prediction using the ARIMA model. In Proceedings—UKSim-AMSS 16th International Conference on Computer Modelling and Simulation, UKSim 2014. https://doi.org/10.1109/UKSim.2014.67
2. Aggarwal, R., & Harper, J. T. (2010). Foreign exchange exposure of “domestic” corporations. Journal of International Money and Finance, 29(8), 1619–1636. https://doi.org/10.1016/j.jimonfin.2010.05.003
3. Akpınar, M., & Yumusak, N. (2013). Forecasting household natural gas consumption with ARIMA model: A case study of removing cycle. 1–6. https://doi.org/10.1109/ICAICT.2013.6722753
4. Anaroua, A., Adhikari, H., & Choi, Y. (2023). Analysis of US Airline Stock Performance using Latent Dirichlet Allocation (LDA). https://doi.org/10.21203/rs.3.rs-3084031/v1
5. Armstrong, J. S. (1985). Long-Range Forecasting: From Crystal Ball to Computer. Wiley. https://archive.org/details/longrangeforecas0002arms
6. Athanasopoulos, G., Poskitt, D. S., & Vahid, F. (2012). Two Canonical VARMA Forms: Scalar Component Models Vis-à-Vis the Echelon Form. Econometric Reviews, 31(1), 60–83. https://doi.org/10.1080/07474938.2011.607088
7. Bhandari, H., Rimal, B., Pokhrel, N. R., Rimal, R., & Dahal, K. (2022). LSTM-SDM: An integrated framework of LSTM implementation for sequential data modeling. Software Impacts, 14. https://doi.org/10.1016/j.simpa.2022.100396
8. Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent dirichlet allocation. J. Mach. Learn. Res., 3(null), 993–1022.
9. Bollen, J., Mao, H., & Zeng, X. (2011). Twitter mood predicts the stock market. Journal of Computational Science, 2(1), 1–8. https://doi.org/10.1016/j.jocs.2010.12.007
10. Bothos, E., Apostolou, D., & Mentzas, G. (2010). Using Social Media to Predict Future Events with Agent-Based Markets. IEEE Intelligent Systems, 25, 50–58. https://doi.org/10.1109/MIS.2010.152
11. Bowman, S. R., Angeli, G., Potts, C., & Manning, C. D. (2015). A large annotated corpus for learning natural language inference. In L. Màrquez, C. Callison-Burch, & J. Su (Eds.), Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing (pp. 632–642). Association for Computational Linguistics. https://doi.org/10.18653/v1/D15-1075
12. Box, G. E., Jenkins, G. M., Reinsel, G. C., & Ljung, G. M. (2015). Time series analysis: Forecasting and control. John Wiley & Sons.
13. Cassel, G. (1916). The Present Situation of the Foreign Exchanges. The Economic Journal, 26(103), 319–323. https://doi.org/10.2307/2221918
14. Chaboud, A., Rime, D., & Sushko, V. (2023). The foreign exchange market. BIS Working Papers. https://www.bis.org/publ/work1094.htm
15. Chen, M., Guo, Z., Abbass, K., & Huang, W. (2022). Analysis of the impact of investor sentiment on stock price using the latent dirichlet allocation topic model. Frontiers in Environmental Science, 10. https://doi.org/10.3389/fenvs.2022.1068398
16. Chen, S., & Hsu, C. (2019). Do stock markets have predictive content for exchange rate movements? Journal of Forecasting, 38(7), 699–713. https://doi.org/10.1002/for.2592
17. Coudert, V., Mignon, V., & Penot, A. (2008). Oil price and the dollar. Energy Studies Review, 15. https://doi.org/10.15173/esr.v15i2.508
18. Crawford, G., & Fratantoni, M. (2003). Assessing the forecasting performance of regime-switching, ARIMA and GARCH models of house prices. Real Estate Economics, 31, 223–243. https://doi.org/10.1111/1540-6229.00064
19. Feng, W., Guan, N., Li, Y., Zhang, X., & Luo, Z. (2017). Audio visual speech recognition with multimodal recurrent neural networks. 2017 International Joint Conference on Neural Networks (IJCNN), 681–688. https://doi.org/10.1109/IJCNN.2017.7965918
20. Feunou, B. (2009). No-Arbitrage VARMA Term Structure Models with Macroeconomic Variables. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.1360864
21. Fischer, T., & Krauss, C. (2018). Deep learning with long short-term memory networks for financial market predictions. European Journal of Operational Research, 270(2), 654–669. https://doi.org/10.1016/j.ejor.2017.11.054
22. Freeman, J., Hays, J., & Stix, H. (2000). Democracy and markets: The case of exchange rates. American Journal of Political Science, 44. https://doi.org/10.2307/2669258
23. García, F., Guijarro, F., Oliver, J., & Tamošiūnienė, R. (2023). Foreign Exchange Forecasting Models: ARIMA and LSTM Comparison. ITISE 2023, 81. https://doi.org/10.3390/engproc2023039081
24. Ghalayini, L. (2013). Modeling and Forecasting the US Dollar/Euro Exchange Rate. International Journal of Economics and Finance, 6(1), p194. https://doi.org/10.5539/ijef.v6n1p194
25. Gillam, L., Ahmad, K., Ahmad, S., Cheng, D., Oliveira, P., & Manomaisupat, P. (2003). Economic News and Stock Market Correlation: A Study of the UK.
26. Giot, P. (2005). Relationships between implied volatility indices and stock index returns. Journal of Portfolio Management, 31(3), 92–100.
27. Glorot, X., Bordes, A., & Bengio, Y. (2010). Deep Sparse Rectifier Neural Networks. 15.
28. Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep learning. MIT press.
29. Guyon, I., & Elisseeff, A. (2003). An introduction to variable and feature selection. The Journal of Machine Learning Research, 3(null), 1157–1182.
30. Hai, Z., Chang, K., Kim, J.-J., & Yang, C. C. (2014). Identifying Features in Opinion Mining via Intrinsic and Extrinsic Domain Relevance. IEEE Transactions on Knowledge and Data Engineering, 26(3), 623–634. https://doi.org/10.1109/TKDE.2013.26
31. Halder, S. (2022). FinBERT-LSTM: Deep Learning based stock price prediction using News Sentiment Analysis (arXiv:2211.07392). arXiv. https://doi.org/10.48550/arXiv.2211.07392
32. Hau, H., & Rey, H. (2006). Exchange Rates, Equity Prices, and Capital Flows. The Review of Financial Studies, 19(1), 273–317.
33. Henckel, T. (2004). Default on Government Debt and Exchange Rate Dynamics. Econometric Society, Econometric Society 2004 Australasian Meetings.
34. Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735–1780.
35. Hodson, T. O. (2022). Root-mean-square error (RMSE) or mean absolute error (MAE): When to use them or not. Geoscientific Model Development, 15(14), 5481–5487. https://doi.org/10.5194/gmd-15-5481-2022
36. How, D. N. T., Loo, C. K., & Sahari, K. S. M. (2016). Behavior recognition for humanoid robots using long short-term memory. International Journal of Advanced Robotic Systems, 13(6), 1729881416663369. https://doi.org/10.1177/1729881416663369
37. Huang, X., Zhang, W., Tang, X., Zhang, M., Surbiryala, J., Iosifidis, V., Liu, Z., & Zhang, J. (2021). LSTM Based Sentiment Analysis for Cryptocurrency Prediction (arXiv:2103.14804). arXiv. https://doi.org/10.48550/arXiv.2103.14804
38. Keynes, J. M. (1923). A Tract on Monetary Reform. Macmillan. https://books.google.com.tw/books?id=8HsuAAAAYAAJ
39. King, M., & Rime, D. (2010). The $4 Trillion Question: What Explains FX Growth Since the 2007 Survey? BIS Quarterly Review.
40. Konchady, M. (2006). Text Mining Application Programming. Charles River Media. https://books.google.com.tw/books?id=m9JQAAAAMAAJ
41. Kumar, M., & M., T. (2014). Forecasting stock index returns using ARIMA-SVM, ARIMA-ANN, and ARIMA-random forest hybrid models. International Journal of Banking, Accounting and Finance, 5, 284. https://doi.org/10.1504/IJBAAF.2014.064307
42. Lin, L., & Kuo, C.-J. (2008). The Relation between Exchange Rate Volatility and Firm Valuation: Polynomial Distributed Lag Model. Journal of Economics and Management, 4, 145–162.
43. Liu, B. (2022). Sentiment analysis and opinion mining. Springer Nature.
44. Mashadihasanli, T. (2022). Stock Market Price Forecasting Using the Arima Model: An Application to Istanbul, Turkiye. Journal of Economic Policy Researches, 9(2), 439–454.
45. Miśkiewicz, J. (2021). Network Analysis of Cross-Correlations on Forex Market during Crises. Globalisation on Forex Market. Entropy, 23, 352. https://doi.org/10.3390/e23030352
46. Naik, D. A., Mythreyan, S., & Seema, S. (2022). Relevance Feature Discovery in Text Mining Using NLP. 2022 3rd International Conference for Emerging Technology (INCET), 1–6. https://doi.org/10.1109/INCET54531.2022.9824807
47. Namouri, H., Jawadi, F., Ftiti, Z., & Hachicha, N. (2018). Threshold effect in the relationship between investor sentiment and stock market returns: A PSTR specification. Applied Economics, 50(5), 559–573. https://doi.org/10.1080/00036846.2017.1335387
48. Nanthiya, D., Gopal, S. B., Balakumar, S., Harisankar, M., & Midhun, S. P. (2023). Gold price prediction using ARIMA model. 2023 2nd International Conference on Vision Towards Emerging Trends in Communication and Networking Technologies (ViTECoN), 1–6. https://doi.org/10.1109/ViTECoN58111.2023.10157017
49. Palangi, H., Deng, L., Shen, Y., Gao, J., He, X., Chen, J., Song, X., & Ward, R. (2016). Deep Sentence Embedding Using Long Short-Term Memory Networks: Analysis and Application to Information Retrieval. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 24(4), 694–707. https://doi.org/10.1109/TASLP.2016.2520371
50. Pang, B., & Lee, L. (2008). Opinion Mining and Sentiment Analysis. Foundations and Trends® in Information Retrieval, 2(1–2), 1–135. https://doi.org/10.1561/1500000011
51. Pornwattanavichai, A., Maneeroj, S., & Boonsiri, S. (2022). BERTFOREX: Cascading Model for Forex Market Forecasting Using Fundamental and Technical Indicator Data Based on BERT. IEEE Access, 10, 23425–23437. https://doi.org/10.1109/ACCESS.2022.3152152
52. Shin, K.-H., Jung, J.-W., Chang, K.-H., Kim, K., Jung, W., Lee, D.-I., & You, C.-H. (2022). Dynamical prediction of two meteorological factors using the deep neural network and the long short-term memory (ΙΙ). Journal of the Korean Physical Society, 80, 1–17. https://doi.org/10.1007/s40042-022-00472-4
53. Shumway, R. H., Stoffer, D. S., & Stoffer, D. S. (2000). Time series analysis and its applications (Vol. 3). Springer.
54. Simionescu, M. (2013). The Use of Varma Models in Forecasting Macroeconomic Indicators. ECONOMICS & SOCIOLOGY, 6(2), 94–102. https://doi.org/10.14254/2071-789X.2013/6-2/9
55. Simon, D. P. (2003). The Nasdaq volatility index during and after the bubble. Journal of Derivatives, 11(2), 9–24.
56. Smales, L. A. (2015). Time-variation in the impact of news sentiment. International Review of Financial Analysis, 37, 40–50. https://doi.org/10.1016/j.irfa.2014.11.019
57. Sousa, M. G., Sakiyama, K., Rodrigues, L. D. S., Moraes, P. H., Fernandes, E. R., & Matsubara, E. T. (2019). BERT for Stock Market Sentiment Analysis. 2019 IEEE 31st International Conference on Tools with Artificial Intelligence (ICTAI), 1597–1601. https://doi.org/10.1109/ICTAI.2019.00231
58. Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. (2014). Dropout: A Simple Way to Prevent Neural Networks from Overfitting. Journal of Machine Learning Research, 1929–1958.
59. Tandel, S. S., Jamadar, A., & Dudugu, S. (2019). A Survey on Text Mining Techniques. 2019 5th International Conference on Advanced Computing & Communication Systems (ICACCS), 1022–1026. https://doi.org/10.1109/ICACCS.2019.8728547
60. Vuong, G. T. H., Nguyen, M. H., & Keung Wong, W. (2022). CBOE volatility index (VIX) and corporate market leverage. Cogent Economics & Finance, 10(1), 2111798. https://doi.org/10.1080/23322039.2022.2111798
61. Whaley, R. E. (1993). Derivatives on market volatility. The Journal of Derivatives, 1(1), 71–84.
62. Y. Ayyappa, B. V. Kumar, S. P. Priya, S. Akhila, T. P. Vardhan Reddy, & S. M. Goush. (2023). Forecasting Equity Prices using LSTM and BERT with Sentiment Analysis. 2023 International Conference on Inventive Computation Technologies (ICICT), 643–648. https://doi.org/10.1109/ICICT57646.2023.10134443
63. Yıldırım, D. C., Toroslu, I. H., & Fiore, U. (2021). Forecasting directional movement of Forex data using LSTM with technical and macroeconomic indicators. Financial Innovation, 7(1), 1. https://doi.org/10.1186/s40854-020-00220-2
64. 周佩親. (n.d.). 探討重大新聞對於匯率之影響-以新台幣兌美元為例 [國立政治大學]. In 國際經營與貿易研究所: Vol. 碩士. 臺灣博碩士論文知識加值系統. https://hdl.handle.net/11296/uy7cs9
65. 曾子維. (2021). 基於新聞情緒建構LSTM神經網路之匯率預測模型與量化交易策略 [國立臺灣師範大學]. In 管理研究所: Vol. 碩士. https://hdl.handle.net/11296/682jx5
66. 李揚生. (2017). 文字探勘與財經新聞: 新聞是否會改變市場行為 [國立中山大學]. In 財務管理學系研究所: Vol. 碩士. 臺灣博碩士論文知識加值系統. https://hdl.handle.net/11296/2px56t
67. 王彥鈞. (2017). 不同市場狀態下新聞情緒的預測能力：以台灣五十指數為例 [國立中央大學]. In 財務金融學系: Vol. 碩士. https://hdl.handle.net/11296/3cea6s
68. 謝人俊 & 林耀傑. (2020). 數位金融環境下銀行經營型態的演變與對策. 中央銀行金融業務檢查處.
69. 陳裕菘, 謝邦昌, 李勝輝, & 陳郁婷. (2014). 運用文字探勘與資料採礦技術建立匯率預測模型－以人民幣兌新台幣為例. Journal of Data Analysis, 9(1). https://doi.org/10.6338/JDA.201402_9(1).0007