Frame-Based Representation of Lexical, Graphical, and Factual Knowledge for Text-to-Scene Generation
Daniel Bauer, Robert Eric Coyne, Owen Rambow · 2012
We outline ongoing work on WordsEye, a text-to-scene generation system. While WordsEye (Coyne and Sproat, 2001) currently recognizes descriptions of simple spatial relations between objects, we are aiming to add support for complex actions, events and states-of-affairs. To this end we use frame-based representations for both asserted and terminological knowledge. The resulting scenes are static spatial arrangements of pre-existing 3D models.