Describing the intangible (affective metadata) with artificial intelligence in television audiovisual archives

Purpose This study aims to explore the integration of affective metadata in television audiovisual archives as a strategy to enrich the representation and retrieval of media memory. It investigates how artificial intelligence can support the automatic detection and classification of emotions in audiovisual content, and examines the technical, documentary and ethical implications of this innovation. Design/methodology/approach The research adopts a qualitative and theoretical approach based on an interdisciplinary literature review. It combines perspectives from audiovisual documentation, semantic analysis and affective computing. The study is complemented by hypothetical application scenarios that illustrate the potential uses and challenges of affective metadata in real-world information systems. Findings The findings indicate that affective metadata significantly enhances the functionality of television archives in three key areas: emotional accessibility, narrative curation and atmosphere-based retrieval. These innovations improve indexing and search processes, while also enabling new practices of mediation and cultural valorization of audiovisual heritage. Originality/value This article contributes to an emerging field by proposing a conceptual framework for documenting emotional dimensions in audiovisual archives. It highlights the transformative potential of affective metadata and artificial intelligence in redefining the cultural, social and professional roles of television archives in the digital era.