Oracle

2/25/2026

Oracle II, Oracle I, a collaboration with RNDR studio (2023 - 2025)

The Oracle series offers a glimpse into the future of data visualization for TU Delft Library's institutional repository. This custom-coded project utilizes machine-learning techniques like (t-SNE) visual clustering, transforming vast amounts of data from academic publications, including student theses and dissertations, into multi-dimensional maps. The algorithm also groups 'neighbours' based on similarities.

Oracle generates unique covers for each publication and encourages users to explore collections using a touch interface. Users can delve into topical and cross-faculty clusters, filter by disciplines, and even time-travel to view the university's oldest theses. In Oracle II, a spatial installation in the Library's Collection Wall, NFC book scanning functionality (to integrate with physical book collections) and scrapped sample images were added. These are not accessible in the online archive version for copyright reasons.

Credits:

Project lead from TU Delft Library: Vincent Cellucci in collaboration with Alice Bodanzky for co-concept development and spatial design

RNDR Studio: Design, Interaction, and co-concept development

Front-end development: Ricardo Matias

Technical support for physical installation: Koen Wösten (TU Delft student assistant) and New Media Centre