Extending and Evaluating a Platform for Story Understanding

We summarize recent developments in our platform for symbolically representing and reasoning over human narratives. The expressive range of the system is bolstered by the infusion of a large library of knowledge frames, including verbs, adjectives, nouns and adverbs, from external linguistic resources. Extensions to the model itself include alternate timelines (imagined states for goals, plans, beliefs and other modalities), hypotheticals, modifiers and connections between instantiated frames such as causality. We describe a corpus collection experiment that evaluates the usability of the graphical encoding interface, and measure the inter-annotator agreement yielded by our novel representation and tool.