本研究旨在藉由文字探勘技術，利用三種辭庫（Financial Sentiment Dictionary、MPQA Opinion Corpus與Opinion Lexicon）量化成財經新聞文章之每日正負面情緒指標，再使用機器學習的方式建立模型（決策樹、梯度提升決策樹、隨機森林、羅吉斯回歸與支援向量機），針對標普500指數進行預測。再根據模型的準確率，與利用模型所給的預測結果，分別以做多和做空的手法進行投資，最後算出個別之報酬率，將對照組的準確率與報酬率與對照組比較，結合投資者與投機者兩種對於報酬與風險截然不同的態度，分析辭庫與模型之特性。最後發現兩點，一，對於股票預測模型而言，準確率並非決定模型優劣的優良指標。二，投資者與投機者最偏好的組合皆為「LM與梯度提升決策樹」之組合，其準確率為59.92%，關於報酬率，該組合之做多報酬率達15.36%，做空報酬率更遠好於對照組的-11.61%，達4.31%，若以一比一的本金同時以做多與做空方式投資，有避險效果，並且可有高達19.67%的報酬率。

In our research, we quantified daily positive and negative sentiment indicators of financial news articles by using three economic sentiment dictionaries (Financial Sentiment Dictionary, MPQA Opinion Corpus and Opinion Lexicon). Then, using these sentiment indicators to build the prediction models (Decision Trees, Gradient Boosting Decision Trees, Random Forests, Logistic Regression and Support Vector Machines) to make predictions for the S&P 500 index.
The main purpose of our research was according to the accuracy of the prediction models and the rates of return of doing long and short by the prediction models to analyze the dictionaries and the model. We compared the accuracy and the rates of return of the experimental group with the control group and determined which dictionary and model investors or speculators prefer. With this research, we have two conclusions. First, accuracy is not a good indicator to select better model for predicting the stock price. Second, the most preferred combination of investors and speculators is LM and GBDT, with an accuracy rate of 59.92%. Additionally, the rate of return for long is 15.36% and for short is 4.31%.

中文文獻
邱培瑄，2017，新聞情緒與黃金之關聯性研究，國立成功大學財務金融研究所在職專班碩士論文。

陳怡安，2017，文本探勘技術之應用-發掘新聞情緒並探討其和恐慌指數之關聯性，國立成功大學財務金融研究所在職專班碩士論文。

英文文獻
Bollen, Johan, Huina Mao, and Xiaojun Zeng. "Twitter mood predicts the stock market." Journal of computational science 2.1 (2011): 1-8.

Chen, Ming-Syan, Jiawei Han, and Philip S. Yu. "Data mining: an overview from a database perspective." IEEE Transactions on Knowledge and data Engineering 8.6 (1996): 866-883.

Christie–David, Rohan, Mukesh Chaudhry, and Timothy W. Koch. "Do macroeconomics news releases affect gold and silver prices?." Journal of Economics and Business 52.5 (2000): 405-421.

Fama, Eugene F., et al. "The adjustment of stock prices to new information." International economic review 10.1 (1969): 1-21.

Fama, Eugene F. "Efficient capital markets: II." The journal of finance 46.5 (1991): 1575-1617.

Fayyard, U. "From data mining to knowledge discovery in databases." AI magazine (1996).

Fayyad, Usama, Gregory Piatetsky-Shapiro, and Padhraic Smyth. "The KDD process for extracting useful knowledge from volumes of data." Communications of the ACM 39.11 (1996): 27-34.

Gillam, Lee, et al. "Economic News and Stock Market Correlation: A Study of the UK Market." Conference on Terminology and Knowledge Engineering. 2002.

Hotho, Andreas, Andreas Nürnberger, and Gerhard Paaß. "A brief survey of text mining." Ldv Forum. Vol. 20. No. 1. 2005.

Nassirtoussi, Arman Khadjeh, et al. "Text mining for market prediction: A systematic review." Expert Systems with Applications 41.16 (2014): 7653-7670.

Qian, Bo, and Khaled Rasheed. "Stock market prediction with multiple classifiers." Applied Intelligence 26.1 (2007): 25-33.

Schumaker, Robert P., and Hsinchun Chen. "Textual analysis of stock market prediction using breaking financial news: The AZFin text system." ACM Transactions on Information Systems (TOIS) 27.2 (2009): 12.

Schumaker, Robert P., et al. "Evaluating sentiment in financial news articles." Decision Support Systems 53.3 (2012): 458-464.

Shapiro, Adam Hale, and Daniel J. Wilson. "What's in the News? A New Economic Indicator." FRBSF Economic Letter 2017 (2017): 10.

Vu, Tien-Thanh, et al. "An experiment in integrating sentiment features for tech stock prediction in twitter." (2012): 23-38.