Bern Data Talk: Managing Video Data - Experiences, Challenges, and Good Practices

Video data is becoming increasingly important across a wide range of disciplines, from social sciences and humanities to psychology and medicine. At the same time, managing video data presents unique challenges. Large file sizes can make storage, transfer, and processing demanding, often requiring specialized infrastructure. Video is also a complex data type that can be difficult to curate, describe, and make findable and reusable in line with the FAIR principles. Researchers must navigate questions of copyright, consent / data protection, publication, and the use of material from social media platforms. In this Data Talk, researchers will share their first-hand experiences in addressing these challenges. Join us, exchange ideas – and don’t miss the subsequent Apéro for inspiring conversations with speakers and colleagues!   Speakers and Topics

Sebastian Flick B.A., David Herrmann M Sc (University of Bern, Data Science Lab) Programmatic Access to University Compute Resources for Video Data The university of Bern provides its members with free access to a high performance computing cluster (UBELIX) and generative AI tools. The challenge we faced in our project at the Data Science Lab was to use these resources programmatically to process video data. Our talk will shine a light on how we manage the different processing steps on client-devices, Server, the university's GPUStack and UBELIX and share good practices for building efficient data pipelines for video processing.   Dr. phil. Selin Oeksuez (University of Bern, Institut für Erziehungswissenschaft) Using Classroom Video Data to Assess Instructional Quality: Opportunities and Challenges Video data provide a valuable methodological approach for assessing instructional quality by enabling the systematic analysis of teaching practices in authentic classroom settings. Drawing on a professional development intervention study, this contribution illustrates how classroom video data were used to assess the quality of teachers’ implementation of two formative assessment strategies, peer and self-assessment, and subsequently transformed into quantitative indicators to allow systematic comparisons between intervention and waitlist control groups. The contribution also addresses methodological and data-related challenges encountered throughout this process, with particular attention to informed consent, data protection, secure data management, and the systematic coding and transformation of complex video data.   Dr. des. Andrea Ferretti (Bern University of Applied Sciences, Bern Academy of the Arts) Making Musical Expression Visible: Managing Sensitive Video Data in Instrumental Music Education Research Video data provide unique opportunities for studying the embodied, multimodal, and situated character of social interaction. At the same time, the richness that makes video analytically valuable also creates particular challenges for data management, data protection, and reuse. Drawing on videographic research into musical expression in instrumental lessons, this talk reflects on focused ethnography and video interaction analysis. It examines how recording decisions shape what can be analysed and how faces, voices, gestures, and musical performances are essential research data and identifying information. Anonymisation may thus protect participants while obscuring the phenomenon under study. The talk discusses differentiated consent, transparent analytical decisions, and graduated access. It argues that responsible management requires balancing scientific value, FAIR principles, and participants’ control over their representation.

When? Tuesday, 27 October, 2026, 4:15 – 5:15 pm, followed by an Apéro for in-depth discussions.

Where? Room 033, University main building, Hochschulstrasse 4