An AI-Based Process for Generating Games from Flat Stories

Abstract

TERENCE is an FP7 ICT European project that is developing an adaptive learning system for supporting poor comprehenders and their educators. Its learning material are books of stories and games. The so-called smart games serve to stimulate the story comprehension. This paper focuses on the analysis of flat stories with a specific annotation language and the generation of smart games from the analysed texts, all done mixing natural language processing and temporal constraintreasoning technologies. It first describes the annotation language, focusing on the key elements for the generation process and stressing which features depends on the requirements of the TERENCE learners. Then it shows how the language is used for annotating stories by the natural language processing module of TERENCE, and how the semantics is provided and is used by the constraint module of TERENCE. Finally, the paper illustrates the generation process of games from the annotation, with example. We conclude commenting on the approach to the automated analysis and extraction of information from stories for specific users and domains, briefly outlining the benefits of the semi-automatic generation process in terms of production costs.