A Psychology-Inspired Approach to Automated Narrative Text Comprehension

ABSTRACT. This paper brings together the theory of argumentation in AI with the psychology of story comprehension to develop a KR\&R framework for narrative text comprehension. The proposed framework is based on a computational interpretation of the general psychological principles of coherency and cognitive economy for text comprehension. Part of a wider study, this paper concentrates on the central problem of integration (or elaboration) of explicit information from the text with implicit (in the reader's mind) common sense world knowledge pertaining to the topic(s) of the story. We show how some of the major challenges for successful story comprehension are closely related to those of NMR, as found in Default Reasoning and Reasoning about Action and Change, and more specifically Knowledge Qualification and Revision. We show how an appropriate formulation through argumentation can give a basis for automating the construction and revision of a coherent comprehension model of a narrative. We report on a first set of empirical experiments designed to harness the implicit world knowledge used by humans reading a story to guide the background knowledge given as input to our framework. We present a prototype system that is used to evaluate the ability of our approach to capture both the majority and the variability of understanding of a story by the human readers in the experiment.