Story-telling maps generated from semantic representations of events

Narratives enable readers to assimilate disparate facts. Accompanying maps can make the narratives even more accessible. As work in computer science has begun to generate stories from low-level event/activity data, there is a need for systems that complement these tools to generate maps illustrating spatial components of these stories. While traditional maps display static spatial relationships, story maps need to not only dynamically display relationships based on the flow of the story but also display character perceptions and intentions. In this work, we study cartographic illustrations of historical battles to design a map generation system for reports produced from a multiplayer battle game log. We then create a story and ask viewers to describe mapped events and rate their own descriptions relative to intended interpretations. Some viewers received training prior to seeing the story, which was shown to be effective, though training may have been unnecessary for certain map types. Self-rating correlated highly with expert ratings, revealing an efficient proxy for expert analysis of map interpretability, a shortcut for determining if training is needed for story-telling maps or other novel visualisation techniques. The study's semantic questions and feedback solicitation demonstrate a process for identifying user-centric improvements to story-telling map design.