Composition of Short Stories Using Book Recommendations

Abstract

A challenge in the domain of creative story generation is generating stories that reflect a consistent theme or genre. We present GREEN, a computational creative (CC) system designed to generate short stories whose genre is customized to a user’s tastes in literature. Given a book title, GREEN uses association rule mining to identify appropriate book recommendations from which to train a long short-term memory model (LSTM) to then produce a new short story. The system forms its intention based on the user’s input, thereby aiming to generate artifacts that will be of value and interest to the user. We report the results of a preliminary survey that serves to demonstrate that as a proof of concept the system shows promise in its ability to achieve its defined intentions.Source: https://github.com/delaneyemoore/GREEN-system