Research On Sentiment Analysis Of Financial Media Information Based On Deep Learning

With the rapid development of the Internet,the financial sector presents a new and prosperous scene.Massive financial media information constantly interacts with investors through the Internet.The investor sentiment affects investors' investment decisions,affects stock price fluctuations,and thus affects the stability of financial markets.This study attempts to conduct sentiment analysis of financial media information and explore the investor sentiment contained therein.In addition,the continuous development of deep learning methods provides "advanced weapons" for this research.Therefore,this study is based on the deep learning method,the sentiment analysis of financial media information,mainly studies the following two parts: the construction of sentiment dictionary,and the sentiment classification of financial news.First of all,since the current sentiment lexicons are mostly general sentiment lexicons,which are not applicable to the financial field,this study uses the word2 vec and constructs a neural network classifier to propose a method for automatically constructing a domain sentiment dictionary,and experiments to prove the method's Effectiveness.Then,this study takes financial news as the research object,based on the LSTM and attention mechanism,proposes a hierarchical attention model to explore the investor sentiment in the news,and experiments prove that the model is better than The statistical method based on sentiment dictionary,traditional machine learning method and general deep learning method also prove that adding attention mechanism can improve the performance of the model.Based on the proof of model performance,this study attempts to explore the correlation between investor sentiment and stock price.The empirical finding that there is a mutual influence between investor sentiment and stock price: the stock price fluctuation of investors in the news is delayed;stock price fluctuations will affect the investor sentiment in the news,and the impact is relatively fast.Finally,summarize the fulltext research and propose the shortcomings and prospects.