Onirica () .interactive was developed as a support to the narrative underlying the body of works of Onirica (). It consists of an interactive, three-dimensional visualisation combining the 28,748 dreams of the banks of the University of Bologna and University of California Santa Cruz at the basis of the work.

Initially conceived as an R&D project, the dream map was originally created for internal use during the development of Onirica (). Its purpose was to help the team orient themselves within the dream databases, which were extremely complex to navigate and comprehend due to their size. The database was navigated through an automated text analysis system with the goal of recognizing patterns, recurring themes and conceptual or semantic connections. After an initial pre-processing session, the text of each dream is analysed by a Machine Learning (ML) model, which associates each of these with a point in a 384-dimensional space: the greater the affinity between the thematic or semantic content of two dreams, the greater their closeness. Through a dimensionality reduction method, the dream space was then transposed into three and two dimensions, generating a point cloud dream map that can be read and interpreted by human eyes.

This system proved to be extremely insightful during the production of the work, which is the reason why it was then revised and improved to be exhibited as an additional piece along Onirica (). Conveying the complexity and magnitude of the research that inspired this body of work turned out to be quite challenging: for this reason, making the vastness of the dataset visible to the public could facilitate a deeper understanding of the work. More than anything else, Onirica () .interactive brings to light the invisible connections that correlate geographically and temporally distant dreams, emphasising similarities and recurring themes.

Onirica () .interactive was developed on Three.js as an application and it directly contains the texts of the dream banks in their original form, not modified or simplified in any way. The user has two main ways to interact with the map: first by typing a keyword on the screen, which triggers the system to highlight all the dreams that contain it. In addition, one can explore the virtual space freely by swiping around, zooming in and out. This way to explore the dataset allows users to easily identify thematic clusters of dreams on the map, stimulating curiosity through a horizontal exploration of the dream banks. For instance, we have located one area dedicated to dreams that focus on teeth, and another one on Superman. To know more about the production process of Onirica () .interactive, here’s a link to our online workshop.

You can visit the artwork at this link.