Controlling inferences by character modeling in story understanding

Abstract

A natural language processing system finally requires understanding the context, and this can be done by the inference. Controlling inferences by tying together the text has been an important issue in story understanding. In this thesis, for story understanding, an inference controlling mechanism through the inference about the story character``s personalities is suggested. The story character``s personalities are inferred from the story events using the behavior interpretation process and represented with Rieger``s character model. The behavior interpretation process considers the knowledge structures such as the affects, the roles, the settings, and the interpersonal relationships in a story to avoid misunderstanding the story.