EMOFIEL: Mapping Emotions of Relationships in a Story

Abstract

We present EMOFIEL, a system that identifies characters and scenes in a story from a fictional narrative summary, generates appropriate scene descriptions, identifies the emotion flow between a given directed pair of story characters in each interaction, and organizes them along the story timeline. These emotions are identified using two emotion modelling approaches: categorical and dimensional emotion models. The generated plots show that in a particular scene, two characters can share multiple emotions together with different intensity. Furthermore, the directionality of the emotion can be captured as well, depending on which character is more dominant in each interaction. EMOFIEL provides a web-based GUI that allows users to query the annotated stories to explore the emotion mapping of a given character pair throughout a given story, and to explore scenes for which a certain emotion peaks.