The paper entitled “Modeling and Prediction of Stock Price with Convolutional Neural Network Based on Blockchain Interactive Information”, which was written by our school’s Professor Li Junfeng (the corresponding author) together with Professor Zhang Wei and master degree candidate Tao Kexin of CUFE School of Information, has been published in the Wireless Communications & Mobile Computing (2020, Article ID: 6686181). Based on the interactive media text data from 2010 to 2015, this research analyzes investor sentiments under the influence of official news and compares the prediction of short-term stock trends with and without sentiment analysis.
The paper finds that how much information is released, disseminated and accepted can have an impact on investor sentiment, which in turn leads to certain volatility in the stock market. However, there are difficulties in analyzing the impact on stock market based on textual content, e.g. strongly interactive information and intermittent disclosure time. To conquer these difficulties, we design a stock price prediction model based on the blockchain architecture for the sentiment analysis of interactive media texts.
The research deals with two aspects mainly: First, a convolutional neural network (CNN)-based sentiment feature extraction model for interactive media texts is constructed to fuse the texts to be classified with their contextual information. Then, we manually label the emotional orientation of interactive media text data and use this as the training set in training the CNN model, effectively improving the accuracy of the interactive textual analysis model. Second, combined with investors’ emotional features under the guidance of official information, the stock price prediction model based on long short-term memory is proposed for discussing the depth and width of the impact of emotional factors. The experiment results show that the accuracy of the model has been improved by incorporating the intervened emotional features, thereby proving that information clarification can have a positive effect on the stock price.
This study provides guidance on how investors and listed companies maximize their respective interests in the market, and provides decision-making solutions for market participants, thus providing important theoretical references and practical guidance for safeguarding investors’ rights and interests, regulating the behavior of listed companies, and optimizing the stability mechanism of the securities market.
