AI-Based Sentiment Analysis for Stock Market Prediction: A Systematic Literature Review

Abstract

The integration of artificial intelligence and natural language processing into financial market prediction has attracted significant research attention over the past decade. This review systematically examines the landscape of AI-driven sentiment analysis techniques applied to stock market movement prediction, covering 87 studies published between 2011 and 2024. A taxonomic framework is proposed to categorize existing approaches along four dimensions: data sources, sentiment extraction techniques, correlation modeling strategies, and evaluation metrics. Quantitative comparisons across lexicon-based, machine learning, deep learning, and large language model paradigms reveal that transformer-based models achieve the highest directional accuracy (up to 65.28%) on standard benchmarks, while lexicon-based methods retain advantages in computational efficiency and interpretability. Key challenges including data noise, temporal decay, and cross-market generalizability are critically assessed, alongside emerging trends in multimodal fusion and explainable AI for financial sentiment.