News Sentiment, Social Media Public Opinion, and Stock Market Volatility: A Review of Multisource Data Forecasting Research

Abstract

The accelerating digitalization of financial information has positioned news sentiment and social media as increasingly influential forces shaping stock market dynamics. This review synthesizes recent interdisciplinary research that leverages multisource data—ranging from mainstream financial newswires to microblogging platforms and retail investor forums—to forecast market volatility. Unlike conventional surveys that concentrate narrowly on algorithmic optimization, we organize our discussion around three interconnected dimensions: the heterogeneous characteristics of sentiment signals across data sources, the theoretical mechanisms that translate public mood into trading behavior, and the practical challenges of integration in predictive modeling. Our analysis reveals that while individual data streams offer complementary predictive power, their fusion encounters persistent obstacles related to temporal misalignment, noise differentials, and interpretability trade-offs. We further identify that much of the existing literature remains preoccupied with short-term forecasting horizons, often neglecting the differential impact of sentiment on volatility regimes across bull and bear markets. This review concludes by outlining a research agenda that emphasizes causal identification strategies, multimodal sentiment decomposition, and adaptive modeling frameworks that can accommodate the rapidly evolving information ecology of modern financial markets.