Context-Aware Sentiment Forecasting via LLM-based Multi-Perspective Role-Playing Agents

2025-05-30

AI Summary

This paper addresses the challenge of modeling emotional evolution in ongoing social media events by introducing *proactive sentiment prediction*: forecasting users’ future sentiment orientations during event progression, moving beyond conventional retrospective sentiment analysis. Methodologically, we propose a novel multi-perspective role-playing framework, wherein LLM-driven agents collaboratively simulate expert analysts, firsthand participants, and neutral observers to perform context-aware, temporally grounded reasoning. Our approach jointly models event-specific contextual cues and fine-grained sentiment dynamics across both micro-level (user-wise) and macro-level (topic-wise) granularities. Extensive experiments demonstrate that our method achieves a 12.6% improvement in F1-score and reduces temporal prediction error by 37.2% over state-of-the-art baselines. These results establish a new paradigm for dynamic, forward-looking sentiment modeling in evolving social media events.

Technology Category

• Cognitive Modeling & Cognitive Systems: Social Cognition And Interaction
• Natural Language Processing: Sentiment Analysis, Stylistic Analysis, and Argument Mining
• Multiagent Systems: Agent-Based Simulation and Emergent Behavior

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Sentiment analysis and opinion mining
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

User sentiment on social media reveals the underlying social trends, crises, and needs. Researchers have analyzed users' past messages to trace the evolution of sentiments and reconstruct sentiment dynamics. However, predicting the imminent sentiment of an ongoing event is rarely studied. In this paper, we address the problem of extbf{sentiment forecasting} on social media to predict the user's future sentiment in response to the development of the event. We extract sentiment-related features to enhance the modeling skill and propose a multi-perspective role-playing framework to simulate the process of human response. Our preliminary results show significant improvement in sentiment forecasting on both microscopic and macroscopic levels.

Problem

• Predicting future user sentiment on social media
• Enhancing sentiment modeling with multi-perspective simulation
• Improving sentiment forecasting accuracy at micro and macro levels

Innovation

• LLM-based multi-perspective role-playing agents
• Sentiment-related feature extraction for modeling
• Simulates human response process dynamically