FSM Formula: An AI-Assisted Framework for Measuring News Article Neutrality

This study proposes the FSM formula, a three-dimensional quantitative framework for assessing news article neutrality. F (Framing Bias) measures whether linguistic framing is unidirectional; S (Source Diversity) measures whether voices are pluralistic; M (AI-based Stakeholder Inference) measures whether relevant groups are systematically omitted. Validated on 18 Taiwanese news articles across five genres (lifestyle, professional, education, finance, politics), findings show that source diversity (S) is a cross-genre structural weakness, and political reporting shows systematic bias as a structural norm. The M module's stakeholder coverage gap calculation is an original contribution not found in existing literature including CHI 2025, AMCIS 2025, and ACL 2025. Dataset available as companion record.