While disaster communication research has predominantly focused on acute crisis phases, how agendas are co-constructed during “milestone moments” of collective remembrance remains understudied. This study addresses this gap by analyzing Japan’s Yahoo! News—a portal aggregator that facilitates the isolation of content-driven responses from the algorithmic and network-structural confounds prevalent on social media. Using content analysis and BERT-based machine learning on 190 articles and 67,563 comments regarding the 10th anniversary of the Great East Japan Earthquake, we identified dual pathways of public engagement. Human-interest framing consistently elicited empathetic, commemorative discourse with high reader approval, functioning as bottom-up emotional risk communication, whereas politically framed coverage generated polarized reactions with significantly lower approval rates. Co-occurrence analysis showed that disaster narratives integrated multiple interconnected themes, while nuclear power and Tokyo Olympics coverage remained isolated. Cross-tabulation confirmed strong content–response correspondence, demonstrating that specific news focuses systematically trigger predictable response patterns. These findings extend agenda-setting theory in two ways: media influence operates through emotional and cognitive frames embedded in content rather than merely through topic salience, and platform architecture functions as a critical moderating variable. The study proposes four complementary roles for disaster coverage beyond information provision: forming emotional memories, facilitating political deliberation, building interactive knowledge, and fostering future-oriented social imagination. These findings offer practical implications for platform-sensitive disaster communication that leverages empathetic framing while mitigating polarization. • BERT-based machine learning helps classify 67,563 disaster-related comments. • Anniversary coverage on Yahoo! News Japan features dramatic, human-interest storytelling. • Human-interest reporting catalyzes bottom-up memory sharing and empathy online. • Political disaster coverage generates polarized reactions with lower approval rates. • Comment sections act as bottom-up hubs for practical knowledge transfer.
Ko Yamada (Sun,) studied this question.