AI tada: Automated indexing of historical television data
The project aims at creating an audio-visual timeline of Dutch television programming by integrating three datasets of Beeld en Geluid. It makes use of television guides (digitized paper collection), metadata of individual television programs (collected in the DAAN-database) and video recordings (per year 2 weeks of 24 hours recordings of all Dutch TV channels) of what was actually broadcast. By combining these datasets, the project increases the accessibility of Beeld en Geluid’s extensive collection and at the same time provides insights into the logics of the medium television and adds historical context to the collection.
To create an audio-visual timeline the project starts from the television recordings (two weeks per year) and connects them to the TV guides and metadata of these weeks. Such a timeline facilitates the design of a user-friendly interface that Beeld & Geluid can implement in its exhibition space to recreate historical mediascapes and make its audio-visual collection more easily accessible.
The project furthermore compares the timelines of the published programming (TV guide) and the actual broadcasts (recordings) to detect possible discrepancies. The results of this comparison is positioned in historical context (differences might be due to unexpected news events) .
Researchers
Ivar Frisch, Alexander Apers, Ying-Kai Dang, Hannah Feldmann, Master’s students
Academic supervisors
Jasmijn van Gorp, Judith Keilbach, Albert Salah, Remco Veltkamp
Grant funding agency; (co-)funding (non-)academic partners
Faculty of Science, Faculty of Humanties, Beeld & Geluid
Publications
van Gorp, J. (2022). Curated playlist: Finding interstitials in a television archive. Web publication/site, CLARIAH. https://mediasuite.clariah.nl/learn/subject-tutorials/curated-playlist-…
Timar, Y., N. Karslioglu, H. Kaya, A.A. Salah, "Feature selection and multimodal fusion for estimating emotions evoked by movie clips," ACM Int. Conf. on Multimedia Retrieval (ICMR), Yokohama, 2018.