Output details

Instantiating interactive narratives from patient education documents

<28> This research revisits previous attempts at generating 3D animations from text, in order to make complex documents easier to understand. Our approach consists in using plan-based representations underpinning interactive narratives as targets for natural language processing, hence providing a higher level of abstraction for semantic mapping. The originality of this method is to optimise the balance between default knowledge, not conveyed by the text, and situations explicitly described. We identify specific reference issues, and explore representative examples within a first interactive narrative prototype dedicated to patient education in surgery. These constitute early results from the FP7 FET project MUSE (FP7-296703).