Viewpoint-based Mining Stock Prices Forecast

Accurate prediction of stock prices is long-standing concern of investors and scholars.There are many factors that affect the stock price movements,and this has brought the stock price forecasting difficult.This paper focuses on the relationship between information and stock prices trying to predict stock price movements by the information change,which identifies the correlation law between stock prices and stock market information from past data and uses it to predict the future price movement.As the deep development of Internet in recent years,people discover that stock market information on Internet is obviously relevant to the stock price,and disclosing the relevance relation becomes the focal point of current research in computing and finance field.This paper introduces a word polarity based opinion mining approach, which extracts the opinion orientation from Internet stock market information,and analyzes the correlation between opinion orientation and stock price movements to predict the future stock price movement.Opinion orientation in stock market information is the expressed position by author whether stock price will raise or drop.Word polarity is the word property of association with the stock price.Word polarity based opinion mining updates the opinion mining data by the recognized word polarity,which implements the opinion extraction on sentence-level,and largely improves the quality of mining data.Neural Network can efficiently predict stock price with its strong capability of approximating any nonlinear system,and the specialty of self-learning and self-adaption.The approach last applies Neural Network to learn the correlation law between stock price and opinion orientation,and uses it to predict the future stock price movement, Experimental result shows that the method achieves good prediction performance,and demonstrates the correlation exists between opinion orientation from Internet stock market information and stock price movement.