Zero-Shot Stance Detection in the Wild: Dynamic Target Generation and Multi-Target Adaptation

TL;DR

This work introduces DGTA, a zero-shot stance detection task in open-world social media where targets are dynamic and potentially multiple per text. It builds a large Chinese multi-domain Weibo dataset with a rigorous cross-validated annotation pipeline and a dual evaluation framework that jointly assesses target identification and stance accuracy, using metrics like $C\text{-}Score$. The authors compare integrated and two-stage fine-tuning of LLMs against prompts and pre-trained baselines, finding that fine-tuned models generally outperform others and that chain-of-thought prompting further enhances performance. The study provides insights into target salience (explicit vs implicit), target quantity effects, and case-specific reasoning capabilities, establishing strong baselines and highlighting directions for future open-world stance detection research.

Abstract

Current stance detection research typically relies on predicting stance based on given targets and text. However, in real-world social media scenarios, targets are neither predefined nor static but rather complex and dynamic. To address this challenge, we propose a novel task: zero-shot stance detection in the wild with Dynamic Target Generation and Multi-Target Adaptation (DGTA), which aims to automatically identify multiple target-stance pairs from text without prior target knowledge. We construct a Chinese social media stance detection dataset and design multi-dimensional evaluation metrics. We explore both integrated and two-stage fine-tuning strategies for large language models (LLMs) and evaluate various baseline models. Experimental results demonstrate that fine-tuned LLMs achieve superior performance on this task: the two-stage fine-tuned Qwen2.5-7B attains the highest comprehensive target recognition score of 66.99%, while the integrated fine-tuned DeepSeek-R1-Distill-Qwen-7B achieves a stance detection F1 score of 79.26%.

Figures (8)

• Figure 1: Real-world examples from the Chinese platform Weibo. (a) Single-target example with one target-stance pair. (b) Multi-target example with multiple target-stance pairs.
• Figure 2: Workflow of dataset construction with collaboration between multiple LLMs and human verification
• Figure 3: Prompt template and example for the integrated fine-tuning strategy
• Figure 4: Prompt template and example for the two-stage fine-tuning strategy
• Figure 5: Three representative cases