The Impact Of News Media On Stock Markets-a Big Data Perspective

Stock market volatility comes from the release,dissemination and absorption of information.With the advance of Web techniques,the influence of Web information on the stock market is becoming increasingly salient with its huge amount and fast dissemination.However,due to the limitation of data acquisition and analytical techniques,the study on the mechanism of Web information percolation and its degree of impact on stock markets is yet to be explored.Previous studies usually regarded news as a whole and explored the impact of all news on the overall stock market,while few studies explored the impact of a certain kind of news on the partly stock market in a more in-depth and detailed way.Besides,previous studies mainly focused on partial of vector-based media information,numerical macroeconomic and microeconomic indicators,and it is of great necessity to study a general intelligent computing framework that combines various market signals in different forms.To fill the research gap in this field,this study tended to studying the news media effect of the stock market from the perspective of big data.In order to obtain abundant internet news data,first,more than 1.1 million news information of 36 mainstream Chinese financial websites from 2015 to 2017 were obtained by the developed targeted distributed network crawler;Second,the advanced natural language processing technology was studied to realize the classification of news topic and the quantification of news emotion;Then,this study systematically and comprehensively explored the news effect of the stock market from three perspectives(the actor,the receiver and the manager)through the traditional financial measurement model.However,the traditional financial econometrics is difficult to capture the complex dynamic process of the real stock market,thus this study further proposed a general deep learning framework that can incorporate complex market factors to study their joint effect on the stock fluctuations.The related research results provided a new perspective and technical skills for the research of classical problems of finance.In addition,this study also provided important theoretical reference and decision support for the practice from three different perspectives of investors' behavior,corporate governance,and stock market regulation.The three contributions of this study are as follows:First,different from the traditional research methods of finance,this study directly analyzed massive financial news data from the perspective of text big data.Using intelligent technology to obtain,organize and analyze text big data,it deeply revealed the relationship between Internet financial news and risk fluctuation of stock market,and the results were able to be closer to the essence of things.Second,exploring the news effect of stock market from multiple perspectives.This study constructed a big data analysis framework to systematically and comprehensively analyze the news media effect of the stock market.Specifically,from three different perspectives of the actor(media),the receiver(companies)and the manager,this study explored the specific performance of the media effect of the stock market under different constraints,namely,heterogeneous news content,the industry of the company,the media behavior of the company's managers.Third,the research on the media effect of stock market based on the deep neural network learning mechanism.This study proposed a general deep learning framework that can incorporate complex market factors to study their joint effect on the stock fluctuations.The study provided a new perspective and technical skills for the research of classical problems of finance,and promote the deep integration of finance and artificial intelligence.