Poster: Translating cross-filtered queries into questions

Complex combinations of coordinated multiple views are increasingly used to design tools for highly interactive visual exploration and analysis of multidimensional data. While complex coordination patterns provide substantial utility through expressive querying, they also exhibit usability problems for users when learning required interaction sequences, recalling past queries, and interpreting visual states. As visual analysis tools grow more sophisticated, there is a growing need to make them more understandable as well. Our long-term goal is to exploit natural language familiarity and literacy to directly facilitate individual and collaborative use of visual analysis tools. In this poster, we present work in progress on an automatically generated query-to-question user interface to translate interactive states during visual analysis into an accompanying visual log of formatted text. Our effort currently focuses on a symmetric and thus relatively simple coordination pattern: cross-filtered views. We describe our current thinking about query-to-question translation in a typical cross-filtered visualization of movies, people, and genres in the Internet Movie Database.