Scholarlink

<p>Accurate predictions in financial markets are crucial for decision-making and risk management. Stock prices and exchange rates are becoming increasingly influenced by public sentiment as expressed on social media and other digital forms of communication, including Twitter and Yahoo Finance. The ever-evolving landscape of global finance necessitates precise forecasting of exchange rates for effective financial planning and risk management. A new method for predicting stock prices that incorporates social-based and financial-based sentiment analysis, along with advanced deep learning algorithms. More specifically, the Recurrent Dual Black-winged Kite Convoluted Graph (RDBKCG) model consists of a Recurrent Dual Convoluted Graph Neural Network (RDCGNN) and Black-winged Kite Algorithm (BKA) to increase the accuracy of predictions. The RDCGNN utilizes both temporal and spatial complexity inherent in stock data, while BKA uses network parameters to accommodate the complexities underlying stock data to attain better performance in stock price forecasting. The procedure creates predictions driven by sentiments sourced in minimal or contextualized essential explanations using explainable AI processes to maximize the understandability of like processes. This included Locally Interpretable Model-agnostic Explanations (LIME) to improve understanding. The methodology approach highlights the increasing significance of sentiment analysis in inducing stock price fluctuations, integrating both social and financial sentiment to attain more accurate and transparent forecasts. The results of the experiment indicate that the suggested method surpasses current techniques, attaining a peak accuracy of 99.4%, a precision of 99.3%, a specificity of 99.2%, a recall of 99.56%, and an F1-score of 99.5%. This provides considerable benefits for stakeholders involved in financial planning and strategy formulation. Overall, the proposed methodology enhances the interpretability of sentiment-driven predictions, providing valuable insights for stakeholders in financial planning and decision-making.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Integration of Explainable AI with Deep Learning for Real-Time Sentiment-Driven Stock Price Prediction

  • 期刊论文

  • 发表日期：

    2025年11月24日

查看全文
  • Sravankumar B,
  • G. Bala Krishna,
  • G. Vishnu Murthy,
  • D. Ramana Kumar,
  • M. Sridevi,
  • M. Varaprasad Rao

摘要

Accurate predictions in financial markets are crucial for decision-making and risk management. Stock prices and exchange rates are becoming increasingly influenced by public sentiment as expressed on social media and other digital forms of communication, including Twitter and Yahoo Finance. The ever-evolving landscape of global finance necessitates precise forecasting of exchange rates for effective financial planning and risk management. A new method for predicting stock prices that incorporates social-based and financial-based sentiment analysis, along with advanced deep learning algorithms. More specifically, the Recurrent Dual Black-winged Kite Convoluted Graph (RDBKCG) model consists of a Recurrent Dual Convoluted Graph Neural Network (RDCGNN) and Black-winged Kite Algorithm (BKA) to increase the accuracy of predictions. The RDCGNN utilizes both temporal and spatial complexity inherent in stock data, while BKA uses network parameters to accommodate the complexities underlying stock data to attain better performance in stock price forecasting. The procedure creates predictions driven by sentiments sourced in minimal or contextualized essential explanations using explainable AI processes to maximize the understandability of like processes. This included Locally Interpretable Model-agnostic Explanations (LIME) to improve understanding. The methodology approach highlights the increasing significance of sentiment analysis in inducing stock price fluctuations, integrating both social and financial sentiment to attain more accurate and transparent forecasts. The results of the experiment indicate that the suggested method surpasses current techniques, attaining a peak accuracy of 99.4%, a precision of 99.3%, a specificity of 99.2%, a recall of 99.56%, and an F1-score of 99.5%. This provides considerable benefits for stakeholders involved in financial planning and strategy formulation. Overall, the proposed methodology enhances the interpretability of sentiment-driven predictions, providing valuable insights for stakeholders in financial planning and decision-making.