# Zero-shot Tweet-Level Stance Detection Enhanced by External Knowledge and Reflective Chain-of-Thought Reasoning

arXiv: https://arxiv.org/abs/2606.26571 · HTML: https://mlsanity.com/paper/2606.26571/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Causal reasoning in LLMs, one of 17 directions in LLM reasoning.
- Builds on 18 of its 54 references that are in the map.
- First posted on arXiv 25 Jun 2026.

## Placed in

- LLM reasoning / [Causal reasoning in LLMs](https://mlsanity.com/lr/causal-reasoning-in-llms/)

## Builds on (most cited first, 18 of 18)

- [Multi-Task Deep Neural Networks for Natural Language Understanding](https://mlsanity.com/paper/1901.11504/) (2019)
- [Zero-Shot Stance Detection: A Dataset and Model using Generalized Topic Representations](https://mlsanity.com/paper/2010.03640/) (2020)
- [Stance Detection with Bidirectional Conditional Encoding](https://mlsanity.com/paper/1606.05464/) (2016)
- [Will-They-Won't-They: A Very Large Dataset for Stance Detection on Twitter](https://mlsanity.com/paper/2005.00388/) (2020)
- [Stance Detection with Collaborative Role-Infused LLM-Based Agents](https://mlsanity.com/paper/2310.10467/) (2023)
- [Cross-Target Stance Classification with Self-Attention Networks](https://mlsanity.com/paper/1805.06593/) (2018)
- [Few-Shot Cross-Lingual Stance Detection with Sentiment-Based Pre-Training](https://mlsanity.com/paper/2109.06050/) (2021)
- [Multi-modal Stance Detection: New Datasets and Model](https://mlsanity.com/paper/2402.14298/) (2024)
- [A Challenge Dataset and Effective Models for Conversational Stance Detection](https://mlsanity.com/paper/2403.11145/) (2024)
- [Cross-target Stance Detection by Exploiting Target Analytical Perspectives](https://mlsanity.com/paper/2401.01761/) (2024)
- [Few-Shot Stance Detection via Target-Aware Prompt Distillation](https://mlsanity.com/paper/2206.13214/) (2022)
- [Few-shot Learning for Cross-Target Stance Detection by Aggregating Multimodal Embeddings](https://mlsanity.com/paper/2301.04535/) (2023)
- [Benchmarking zero-shot stance detection with FlanT5-XXL: Insights from training data, prompting, and decoding strategies into its near-SoTA performance](https://mlsanity.com/paper/2403.00236/) (2024)
- [Stanceformer: Target-Aware Transformer for Stance Detection](https://mlsanity.com/paper/2410.07083/) (2024)
- [CoVaxNet: An Online-Offline Data Repository for COVID-19 Vaccine Hesitancy Research](https://mlsanity.com/paper/2207.01505/) (2022)
- [Enhancing Stance Classification on Social Media Using Quantified Moral Foundations](https://mlsanity.com/paper/2310.09848/) (2023)
- [Examining Temporalities on Stance Detection towards COVID-19 Vaccination](https://mlsanity.com/paper/2304.04806/) (2023)
- [A Weakly Supervised Approach for Classifying Stance in Twitter Replies](https://mlsanity.com/paper/2103.07098/) (2021)

## Abstract (arXiv)

Zero-shot tweet-level stance detection confronts two primary challenges: (1) mitigating the context sparsity inherent in short texts, and (2) establishing the relevance between implicit targets and textual content. While existing methods primarily focus on incorporating external knowledge, they neglect the intrinsic semantic cues embedded within key intra-textual entities. Furthermore, current models exhibit limited capability in determining the relevance of unseen targets to the given text, thereby struggling to differentiate between "neutral" and "irrelevant" stance labels. To address these issues, we first construct a four-class, multi-topic Japanese tweet dataset. To our knowledge, this is the first Japanese tweet-level dataset for stance detection. We then propose KIRP, a zero-shot stance detection framework. It integrates external knowledge with entity reorganization for data augmentation and employs prompt chaining for reasoning. Specifically, the framework incorporates knowledge graphs to supplement and reorganize key textual entities, while reflective Chain-of-Thought (CoT) reasoning extracts and validates implicit targets. To better distinguish "neutral" from "irrelevant" labels, we adopt stance-aware contrastive learning to capture discriminative features and design a three-layer iterative prototype network for fine-grained classification. Experimental results on SemEval-2016, WT-WT, and KIRP-D show that KIRP achieves state-of-the-art performance. KIRP obtains F1 scores of 84.05% (three-class) on SemEval-2016, and 84.99% and 79.18% (four-class) on WT-WT and KIRP-D, respectively.