Journalism-Guided Agentic In-Context Learning for News Stance Detection

2025-07-15

AI Summary

In online news recommendation, personalized systems exacerbate filter bubbles and political polarization due to insufficient viewpoint diversity. Existing stance detection methods are primarily designed for short texts and high-resource languages, rendering them inadequate for fine-grained stance analysis of lengthy Korean news articles. To address this gap, we introduce K-News-Stance—the first benchmark dataset for stance detection in long-form Korean news—and propose JoA-ICL, a novel agent-based in-context learning framework grounded in news structure. JoA-ICL identifies key structural segments (e.g., leads, quotations) to enable paragraph-level stance inference and aggregation, integrating large language model agents, structure-aware prompting, and multi-granularity annotations. Experiments demonstrate that JoA-ICL significantly outperforms state-of-the-art baselines. Case studies confirm its effectiveness in uncovering media bias and enhancing viewpoint diversity in news recommendation systems.

Technology Category

• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Machine Learning: Learning Preferences or Rankings
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Web Mining and Content Analysis: Sentiment analysis and opinion mining

Abstract

As online news consumption grows, personalized recommendation systems have become integral to digital journalism. However, these systems risk reinforcing filter bubbles and political polarization by failing to incorporate diverse perspectives. Stance detection -- identifying a text's position on a target -- can help mitigate this by enabling viewpoint-aware recommendations and data-driven analyses of media bias. Yet, existing stance detection research remains largely limited to short texts and high-resource languages. To address these gaps, we introduce extsc{K-News-Stance}, the first Korean dataset for article-level stance detection, comprising 2,000 news articles with article-level and 19,650 segment-level stance annotations across 47 societal issues. We also propose extsc{JoA-ICL}, a extbf{Jo}urnalism-guided extbf{A}gentic extbf{I}n- extbf{C}ontext extbf{L}earning framework that employs a language model agent to predict the stances of key structural segments (e.g., leads, quotes), which are then aggregated to infer the overall article stance. Experiments show that extsc{JoA-ICL} outperforms existing stance detection methods, highlighting the benefits of segment-level agency in capturing the overall position of long-form news articles. Two case studies further demonstrate its broader utility in promoting viewpoint diversity in news recommendations and uncovering patterns of media bias.

Problem

• Detecting article-level stance in Korean news to mitigate filter bubbles
• Addressing limitations of stance detection in long texts and non-English languages
• Improving news recommendations by analyzing segment-level stances in articles

Innovation

• Korean dataset for article-level stance detection
• Journalism-guided agentic in-context learning framework
• Segment-level agency for stance aggregation