A New Sentiment Analysis Method With News Readers’ Attitude And Heat Degree

All the investors hope to earn more by analyzing the market information.With the traditional media becoming less popular,internet media has becoming the main channel for information.People can get immediate information regardless of the time and space.At the same time,people can spread their own opinions on the internet.Readers can figure out what they are interested in and ignore what they are not.However,the information on the internet is much more huge than people can handle.As a result,people admit the value of information on internet but they can not use it efficiently.From one point of view,news media helps the investors to get information.From another part,news media is influencing the investors feelings as well.It is valuable to analyze the sentiment in news media so as to know more about how people make decision and even more how to predict the stock market.The aim of this research is to take use of the news on Internet more effectively.This research had built three kind of sentiment dictionaries,namely the high-frequency sentiment dictionary,the psychology dictionary and the domain dictionary,using three kind of sentiment calculating methods,namely the tf-idf method,the heat degree method and the readers’ attitude method to analyze the sentiment of news.After that,support vector machine was used to build a prediction model.Results showed that the readers’attitude and heat degree improved the accuracy of prediction.The high frequency dictionary achieved the best accuracy of prediction.