An Annotated Corpus of Stories and Gestures for a Robotic Storyteller

Abstract

This paper describes a system which has been used to automatically create an annotated corpus of stories. This corpus can be exploited by a robotic storyteller. In a first phase stories were retrieved from a repository. The textual content of each story was semantically analyzed in order to locate meaningful terms in the sentences that can be accompanied by proper gestures and possible emotions that can be expressed. Once the whole story is properly labelled, it can be dramatized by a robot storyteller. We discuss a demonstration of the use of this approach with NarRob, a storyteller robot that we have designed within our institute.