Stance Detection with Collaborative Role-Infused LLM-Based Agents

TL;DR

COLA introduces a three-stage stance-detection framework that orchestrates collaborative role-infused LLM agents (Linguistic Expert, Domain Specialist, Social Media Veteran) to achieve high-accuracy zero-shot performance without annotated data. The reasoning-enhanced debating stage enables explicit linking of text features to stances, while a final judger consolidates evidence for a robust stance label; ablations confirm each component’s contribution. Across SEM16, P-Stance, and VAST, COLA surpasses zero-shot baselines and rivals in-target methods, and demonstrates explainability through JSON-formatted rationales and quantitative improvements when explanations aid back-models. The framework further proves versatile by extending to aspect-based sentiment analysis and persuasion prediction, underscoring its applicability to broad web and social media text analysis, with ethical and practical considerations discussed for real-world deployment.

Abstract

Stance detection automatically detects the stance in a text towards a target, vital for content analysis in web and social media research. Despite their promising capabilities, LLMs encounter challenges when directly applied to stance detection. First, stance detection demands multi-aspect knowledge, from deciphering event-related terminologies to understanding the expression styles in social media platforms. Second, stance detection requires advanced reasoning to infer authors' implicit viewpoints, as stance are often subtly embedded rather than overtly stated in the text. To address these challenges, we design a three-stage framework COLA (short for Collaborative rOle-infused LLM-based Agents) in which LLMs are designated distinct roles, creating a collaborative system where each role contributes uniquely. Initially, in the multidimensional text analysis stage, we configure the LLMs to act as a linguistic expert, a domain specialist, and a social media veteran to get a multifaceted analysis of texts, thus overcoming the first challenge. Next, in the reasoning-enhanced debating stage, for each potential stance, we designate a specific LLM-based agent to advocate for it, guiding the LLM to detect logical connections between text features and stance, tackling the second challenge. Finally, in the stance conclusion stage, a final decision maker agent consolidates prior insights to determine the stance. Our approach avoids extra annotated data and model training and is highly usable. We achieve state-of-the-art performance across multiple datasets. Ablation studies validate the effectiveness of each design role in handling stance detection. Further experiments have demonstrated the explainability and the versatility of our approach. Our approach excels in usability, accuracy, effectiveness, explainability and versatility, highlighting its value.

Figures (3)

• Figure 1: Illustration of the challenges of stance detection.
• Figure 2: Architecture of our proposed COLA.
• Figure 3: Cases of explainations generated by our approach.