2015. Computational text artwork / software project. Presented at the Print Screen Festival, Tel Aviv.
Word2Dream is a computational text work based on Word2Vec. A user supplies an initial text, from which the system identifies significant words and generates related terms in a semantic vector space. The outputs can then serve as new inputs, producing an evolving chain of associations that moves progressively away from the starting text.
The work treats this recursive traversal as a form of computational “dreaming”: rather than simply analysing or representing a text, the system explores an associative space learned from statistical relationships in language.
Hadas, Eran, and Eyal Gruss. 2015. Word2Dream: A Reader's Companion. Computational text artwork / software project.
Cheesman, Tom, Kevin Flanagan, Stephan Thiel, Jan Rybicki, Robert S. Laramee, Jonathan Hope, and Avraham Roos. “Multi-Retranslation Corpora: Visibility, Variation, Value, and Virtue.” Digital Scholarship in the Humanities, 32(4), 739–760. https://doi.org/10.1093/llc/fqw027