arXiv:2606.06022v1 Announce Type: new Abstract: Stance detection on social media is challenging due to short, noisy, and context-dependent language. While large language models (LLMs) show zero-shot generalization, they are typically prompted without contextual information, which limits their ability to interpret ambiguous posts. In this work, we systematically investigate the impact of incorporating real-world (e.g., user biographies), derived (e.g., political party), and LLM-generated (e.g., target descriptions) contextual features into zero-shot prompting for stance detection on Twitter. Ou
The increasing sophistication of LLMs and the pervasive use of social media necessitate more advanced methods for interpreting public sentiment and preventing misuse.
Improving stance detection on social media through contextualized prompting enhances the accuracy and utility of AI in understanding potentially influential narratives and mitigating misinformation.
LLMs can now more effectively interpret nuanced and ambiguous social media content by incorporating real-world, derived, and internally generated contextual features.
- · Social Media Platforms
- · Information Intelligence Firms
- · AI/ML Researchers
- · Public Opinion Analysts
- · Misinformation Propagators
- · Uncontextualized AI Models
More accurate automated identification of stance on social media content. This can lead to improved content moderation.
Better understanding of public sentiment on controversial topics, informing policy-making and public relations strategies.
Enhanced ability to predict and potentially influence narrative flows, impacting societal discourse and political outcomes.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL