Research On The System Of Book Recommendation In Universities

Nowadays,with the growth of the Internet,a great deal of information is spread by many listed companies on various media.Most investors analyze the information while making investment strategies.However,most investors when buying stocks are not irrational.This is because they do not have access to the true information,lack theoretical knowledge associated with investment and are easily influenced by the market.It has been found that the trading frequency of the stock market will to a large extent,be influenced by various factors such as market emotional effects and the irrational behaviors of shareholders.Moreover,the market uncertainty caused by various types of emergencies will deepen the panic of Internet users.In this case,a new challenge appears in the face of the traditional prediction and analysis technology of the stock business.In recent years,many researchers have found that there is a strong correlation between the market sentiment mapped by the Internet public opinion and the trading frequency of the stock market.The analysis method of the market sentiment trend which is reflected in its inner workings and the Internet public opinion is a key issue in the stock market forecasting technology that needs to be strengthened in analysis and research.The work can be mainly divided into three parts:In the first part,the comments of Vanke A Share(sz000002)will be analyzed with the Chinese text analysis technology,and the learning model of text sentiment polarity analysis with the basis of machine learning will be studied.After processing textual data through ways like word segmentation using the idea,using the emotions in the stock evaluation will be classified as positive ones or negative ones by taking advantage of idea of supervised learning in machine learning,using Word2 vec model to train feature vectors as well as using SVM according to the results of training.The experimental results showed that the accuracy rate of the classification is 88%.In the second part,after gaining the results of the sentiment analysis,the researchers will quantify the results,getting the network public opinion rate and the weighting network public opinion generated by using the classification of weights of comment number.In the third part,researches propose a BP neural network model based on the Internet public opinion,weighted network public opinion and stock technical indicators.By predicting the closing price,the forecast accuracy of the stock price and the upward and downward trend is improved.The empirical analysis results show that the Mean Square Error(MSE)is 0.38182404,The Trend Accuracy(TAR)is 80.34%,and the Absolute Error(AE)is 0.748,all of which are much better than the BP neural network model based only on stock technical indicators only.The prediction result is thus much better.Apart from the trend accuracy that is as high as 85.60%,the indexes in the BP neural network prediction model based on weighting network public opinion value are not ideal.The simulation results show that the BP neural network model with the addition of network public opinion and weighting network public opinion significantly improves both the prediction accuracy and fluctuation rate of the stock price.Overall,it is an effective mode lto forecast stock prices and trends.