In addressing the challenge of enhancing stock price prediction accuracy,this paper proposes a method that employs multi-feature nested attention fusion of stock prices and emotions.It mitigates the negative impacts of interference items on emotional judgment,eliminating noise and rumors.Firstly,features are selected and the relevant indicators are constructed using stock market and emotional text series.Secondly,the traditional time series prediction model ARIMA is utilized to generate a prediction residual sequence,addressing long-term trends and seasonal changes in the stock price time series and capturing random fluctuations.The DFAOA-BERT model is employed to extract emotional text features,construct a sentiment analysis model,and evaluate or classify comments based on sentiment.Finally,the sentiment polarity discrimination results from stock reviews are input into the stock market prediction model,forming a multi-feature nested attention fusion stock price prediction model.This model explores the relationship between investor sentiment and stock returns and prices,aiming to enhance the accuracy of predicting stock price trends.Experimental results indicate that the proposed method effectively improves the accuracy of predicting stock price trends.
NETL
NSTL
万方数据
