Conveying the Predicted Future to Users: A Case Study of Story Plot Prediction

TL;DR

This paper presents a system that generates short descriptions predicting story plots to assist writers, demonstrating that a frame-enhanced GPT-2 produces the most consistent and storiable summaries, with potential to support creative writing.

Contribution

It introduces a novel approach using frame-enhanced GPT-2 for story plot prediction, evaluated through user studies for writing support.

Findings

FGPT-2 outputs are rated most consistent and storiable.

FGPT-2's descriptions outperform some human-written snippets.

Machine-generated plots positively influence the writing process.

Abstract

Creative writing is hard: Novelists struggle with writer's block daily. While automatic story generation has advanced recently, it is treated as a "toy task" for advancing artificial intelligence rather than helping people. In this paper, we create a system that produces a short description that narrates a predicted plot using existing story generation approaches. Our goal is to assist writers in crafting a consistent and compelling story arc. We conducted experiments on Amazon Mechanical Turk (AMT) to examine the quality of the generated story plots in terms of consistency and storiability. The results show that short descriptions produced by our frame-enhanced GPT-2 (FGPT-2) were rated as the most consistent and storiable among all models; FGPT-2's outputs even beat some random story snippets written by humans. Next, we conducted a preliminary user study using a story continuation task where AMT workers were given access to machine-generated story plots and asked to write a follow-up story. FGPT-2 could positively affect the writing process, though people favor other baselines more. Our study shed some light on the possibilities of future creative writing support systems beyond the scope of completing sentences. Our code is available at: https://github.com/appleternity/Story-Plot-Generation.