User adapted natural language discourse generation for validation of conceptual models

Hercules Dalianis

Abstract

The support systems for conceptual modeling of today lack natural language feedback. This thesis argues for the need of a user adapted natural language generation as a validation tool, and also as a support tool during execution of a conceptual model. Based on this conclusion a natural language generation system is suggested. Various appropriate discourses are then proposed in this thesis and are analyzed with Hobbs' coherence relations. The proposed texts could be analysed with fewer coherence relations than the whole theory contains. A discourse grammar is generalized from the analyzed discourses. The thesis continues with an overview of the state of the art of the natural language generation technique. A discussion is carried out for how to control the generation. The suggestion is that the user's question together with a user model and various features of the grammar will control the generation. To conclude the thesis a support system based on the natural generation techniques of today and on previous working systems constructed by the author is suggested.