Text Opinion Mining for Stock Market Prediction using News Reports of Securities in Taiwan

In this paper, we introduce a method to predict stock price by mining text opinions in news reports of securities. In Taiwan, the populace investors usually read/watch the news reports of securities before doing their stock investments. It motivated us to analyze the relations between the changes of stock prices and the opinions in news reports of securities. Our method consists of carrying out the text processing of news report, describe the features, extracting the expressed opinions, and predicting stock price tendency. We use an association rule mining method to analyze the opinions shown in news reports and the changes of stock prices. Our method produces some interesting rules to predict stock prices. Based on the obtained results, we find that the opinions in news reports usually have a great influence on the changes of stock prices. However, not all opinions are equal. Researches can aim at identifying these relationships for the populace investors in Taiwan to predict stock prices.