A Computational Model of Inferencing in Narrative

Recent work in the area of interactive narrative has sought to develop systems that automatically produce experiences for a user that are understood as stories. Much of this work, however, has focused on the structural aspects of narrative rather than the process of narrative comprehension undertaken by users. Motivated by approaches in natural language discourse generation where explicit models of a reader’s mental state are often used to select and organize content in multisentential text, the work described here seeks to build an explicit model of a reader’s inferencing process when reading (or participating in) a narrative. In this paper, we present a method for generating causal and intentional inferences, in the form of sequences of events, from a narrative discourse. We define a representation for the discourse, the sequence of discourse content, and show how it may be translated to a story representation, the reader’s plan. We define cognitive criteria of necessitated inferences with regards to these representations, and show how a partial order planner can determine which inferences are enabled. The inference generation is motivated by findings in cognitive studies of discourse processing, and we provide support for their online generation by readers in a pilot study.