Qurator

A research project cluster between 10 Berlin-Brandenburg based companies explored the information management challenges for curators, media managers and cultural institutions in the world of ever-increasing volume of media content. As an early adopter of machine learning and AI techniques, the research project brought together a broad consortium to think new ways to help organise and manage information.

Our focus at ART+COM was to think about the challenges in the context of knowledge workers when creating exhibitions and stories in public space. The work builds on the research results of its predecessor project, Digital Curation Technologies, to develop innovative technologies for exhibition curation based on Artificial Intelligence. An interdisciplinary and user-centred design approach puts the focus on the usability.

We developed early prototypes to facilitate research of not only text content, but images, videos, 3D models, and spoken language. The approach was to be able to incorporate knowledge from external sources to enable a semi-automated generation and classification of ontologies.

The resulting ideas leveraging early form of RAG — vector optimisation, action detection, pix2pix algorithm and UMAP dimension reduction — were instrumental in a number of commercial exhibits we have built since.