Conceptual
Browse Instruct Curate

Neutralizing Partisan Media Bias via Emotional-Fingerprint Summarization

How to measure and remove political slant from news text by treating partisan bias as an 'emotional fingerprint' — a signature pattern of Valence-Arousal-Dominance affect that differs systematically across left, centre, and right outlets. A neutralizing abstractive summarizer then explicitly targets those affective dimensions to produce coherent multi-source summaries whose emotional profile is measurably closer to neutral, shifting bias mitigation from political labels to the emotional encoding of language.