Associate creates an experimental and poetic digital documentation from the various discussions that take place within the Untitled events. Associate is at the same time an independent artwork and experimental documentation of the festival and the online events.
The recorded discussions in Untitled meetings and festival events are processed through a machine learning algorithm. This process picks up parts of sentences and individual words that accumulate and mix in the ever increasing database. The algorithm forms new connections between various textual elements based on machine learning models that analyse the context and statistical properties of individual words and phrases.
The machine learning model aims to bring light to emerging and new connections between textual nodes, and reflect on various alternative meanings and paths derived from the language used within Untitled discussions.
Images: Otso Havanto / Associate.
Privacy and anonymisation of data is a core function of the artwork. All the conversations recorded in Untitled events that are used for Associate, are processed in a way that it is impossible to identify an individual speaker.
The first version of the artwork is displayed online on its own website. This version creates new textual interpretation clusters and poetic variations from the Untitled discussions, ultimately aiming to foster the creation of Untitled’s own unique discourse engine.
Associate uses a statistical machine learning model that has been trained by the Common Crawl dataset, and by the OntoNotes source material from the University of Pennsylvania. The work accumulates ever expanding textual material from the Untitled discussions and produces novel statistical vectors between the various meanings of words and concepts. Over time the work trains a model that is unique to the language used by the Untitled community and platform.
The artwork aims to encourage us to use big data as a tool and resource for the various communities, movements, and other non-commercial organizations, that aim to create a more just and fair future society for all.
The artwork can be found on the following link: associate.associates