Modeling Tension in Stories via Commonsense Reasoning and Emotional Word Embeddings

Abstract

Dramatic tension is crucial for generating interesting stories. This paper aims to model dramatic tension from a story text using neural commonsense-reasoning language models and emotional word embeddings. We also propose a method of converting a categorical emotion word into a numerical value. The evaluation results using human-annotated stories demonstrate that our proposed method is promising in predicting tension development in a story.

Tasks

Reproductions