SLANG (System for Learned Artificial Narrative Generation) is a project aimed at narrative generation, the design of an algorithm capable of building a fiction narrative from scratch. Most importantly, a narrative must be logically consistent and evolve over time, and ideally be sufficiently complex so as to resemble human level writing. This would require not only a baseline level of natural language understanding but also the ability to a construct high-level abstract storyline and timeline of events, and be able to recursively fill in the details to output a final narrative. RNNs, LSTM RNNs in particular, have shown promise in natural language generation, and this project seeks to build on top of this existing literature with the application of generative adversarial networks or variational autoencoders.
currently no documents uploaded for this project