Representing Story Plans in SUMO

Abstract

Automatic story generation systems require a body of commonsense knowledge about the basic relationships between concepts we find everyday in our world in order to produce interesting narratives that describe human actions and world events. This paper presents an ongoing work that investigates the use of Suggested Upper Merged Ontology (SUMO) to represent storytelling knowledge and its inference engine Sigma to query actions and events that may take place in the story to be generated. The resulting story plan (fabula) is also represented in SUMO, allowing for a single story representation to be realized in various human languages. 1