Question Answering in the Context of Stories Generated by Computers

The QUEST model of question answering is a computational-cognitive model designed to describe a person’s mental narrative comprehension process, which predicts behavioral responses to openended questions about the narrative. The model is based around a QUEST knowledge structure, a directed graph that captures the story’s events and the causal and goal-oriented relationships between them. We present a family of computational methods to transform an automated planning narrative knowledge representation that has been successfully used for computational narrative generation, to a QUEST knowledge structure. We then report an experiment to test the hypothesis that these plan-driven knowledge structures have a meaningful relationship with a story consumer’s comprehension of a narrative. The results help advance a research agenda that uses cognitive states as the target of computationally generated narratives.