A Knowledge Representation that Models Memory in Narrative Comprehension

We present work toward computationally defining a model of narrative comprehension vis-à-vis memory of narrative events, via an automated planning knowledge representation, capable of being used in a narrative generation context. There has been much recent research on computationally analyzing and generating narratives (e.g. Mani 2012). Key to these efforts is the modeling of the mind as it makes sense of stories. As people perceive narrative, their story comprehension faculties are active in the projection of a fictional world (Gerrig 2013), such that the story context in which they are embedded plays a key role in how they expect the future of the narrative to unfold. Authors accordingly design stories to affect their audience in specific ways (Bordwell 1989). A generative computational model of narrative must go beyond story structure, because the fundamental design criteria for a narrative artifact rest in the cognitive and affective responses they prompt in their human consumers. In this paper, we present work toward a computational model of narrative, which begins to account for the human consumer by modeling the person’s memory for previously experienced narrative events relative to the most recently experienced event of the same narrative.