Major goals achieved in the this 4.25-year innovation action include:
In collaboration with peer cultural heritage organizations, the requirements of advanced digitization were assessed, resulting in the creation of a comprehensive Tool Kit intended to inform the digitization of film materials and related non-filmic archival materials. The VHH Team built on working relationships with numerous archives, libraries, and memory institutions holding relevant materials in the USA, UK, Russia, Belarus, Ukraine, , Latvia, and Estonia. Rich data and metadata collections of primary documents on the liberation of concentration camps and the discovery of other atrocity sites (films, photographs, text documents, oral history interviews) and related popular culture products (films, artworks, graphic novels, video games, internet memes) were curated, digitized, and aggregated.
Key curatorial concepts enabling curated access and interactive engagement with these digitized or born-digital cultural heritage assets on a new type of Web platform were discussed, defined, elaborated, and published in numerous deliverable reports.
Strategies for the aggregation of legacy metadata on cultural heritage assets were developed. The VHH Team developed and implemented a standards-compatible metadata schema allowing for the preservation of data in a way that keeps the data faithful to its source. Comprehensive sets of controlled vocabularies were created for the rich and interdisciplinary time-based and non-time-based description of the cultural heritage assets and their contents. A Taxonomy of Relations was developed and used both as an analytical theory of visual relations and as a practical annotation model.
The methodological foundations for the automatic analysis of digitized films and texts were established. Automatic film analysis tools for overscan detection, shot boundary detection, shot type classification, camera movements classification, object detection and tracking, and relation detection were developed and implemented. A human-in-the-loop OCR pipeline for transforming digital copies of text documents into machine-readable text (BOW) was created. A prototype of a radically new way of linking, exploring, and searching digital text archives backed by semantic and contextualized search (ArchAIvist) was designed. All automatic analysis tools were published open source.
The requirements of the Visual History of the Holocaust Media Management and Search Infrastructure (VHH-MMSI, aka the VHH platform) were elicited and continuously updated in an agile approach, the system architecture was defined and implemented, hardware and software were set up, back end and front end were created, expanded, and tested. A new digital film player was developed and and implemented in the VHH-MMSI. Large sets of data and metadata were imported, including the recovered and verified legacy metadata of the Cinematography of the Holocaust.
The project’s objectives, research questions, findings, and prospected results were disseminated through numerous articles and presentations at academic conferences and workshops. Three international academic conferences were organized and hosted by the project in Frankfurt, Jerusalem, and Paris. A new project website was launched after year 1, disseminating and communicating the activities and results of the project to experts and the broad public. Stakeholder gains and pains were analyzed, exploitable results identified, and exploitation roadmaps defined.