We presented our progress in the projects TIB-AV-A and FakeNarratives in two presentations at the Society for Cognitive Studies of the Moving Image Conference (SCSMI) 2024 in Budapest, Hungary. It was nice meeting film scholars to further develop our approaches for film and news analysis.

Eric Müller-Budack, Matthias Springstein, Margret Plank, Julian Sittel, Roman Mauer, Oksana Bulgakowa, Manuel Burghardt, John Bateman, Ralph Ewerth (2024). TIB AV Analytics: A Computational Tool for Film and Video Analysis. Society for Cognitive Studies of the Moving Image Conference, SCSMI, Budapest, Hungary, June 5-8, 2024.

Abstract

In this presentation, we provide an overview of current computational tools and methods for film and video analysis. After briefly contextualizing this direction of development against the evolution of methods in this field, we set out theoretical foundations for empirical video analyses that are fully responsive to the nature of film as an expressive medium. As we focus on using state-of-the-art information extraction methods (typically based on deep learning), we also provide an overview of the types of information that can already be extracted with these methods. Specifically, we introduce a web-based tool for systematic film and video analysis, called TIB AV-Analytics (TIB-AV-A). We show the wide range of analysis methods provided by TIB-AV-A and demonstrate how it can be used to support video analysis. Finally, we conclude by summarizing the current state of available tools and methods for computational video analysis and outline some challenges that lie ahead.

Chiao-I Tseng, John A. Bateman, Leandra Thiele, Ralph Ewerth, Eric Müller-Budack, Gullal Cheema, Manuel Burghardt, Bernhard Liebl (2024). The search for filmic narrative strategies in audiovisual news reporting: a progress report. Society for Cognitive Studies of the Moving Image Conference, SCSMI, Budapest, Hungary, June 5-8, 2024.

Abstract

Audiovisual news reporting is now documented to involve many filmic techniques that bring news reporting ever closer to audiovisual storytelling. At SCSMI-2022 we introduced our FakeNarratives project, which undertakes a contrastive cataloguing of filmic narrative strategies in both mainstream and alternative news media to support the location of potentially problematic messaging. We now discuss the progress that has been made towards automating the recognition of filmic structures using diverse computational techniques for audiovisual processing. Results are maintained in a searchable richly annotated graph structure, allowing us to define narrative patterns in terms of formal combinations of filmic features present in the graph. By these means, we increase the scale of data on which filmic narrative patterns can be derived, empirically validated, and productively visualised. As the analytic framework is oriented to audiovisual material in general, we also show how aspects of the account may contribute to film research more broadly.