Conceptual Authoring generalizes the technique we have referred to as WYSIWYM, or What You See Is What You Meant.
WYSIWYM arose from two NLG projects (DRAFTER and GIST) undertaken in the early 1990s at University of Brighton. In these projects we needed a method for encoding the semantic content of instructional materials in a logical formalism, from which equivalent texts could be generated in several natural languages. After trying various graphical methods, we came up with the idea of using NLG as a means of obtaining its own input. Instead of manipulating an unfamiliar graphical presentation such as an entity-relationship diagram, the author performs logically equivalent operations by direct manipulation of a ‘feedback text’ generated by the system.
So far, the WYSIWYM label has been applied almost exclusively to systems we have developed, either at University of Brighton or the Open University, in which an author edits assertional material by interacting with a feedback text. We are introducing ‘Conceptual Authoring’ as a more general term for this area of research, for several reasons: