The University of Edinburgh -
Division of Informatics
Forrest Hill & 80 South Bridge

PhD Thesis #9213

Title:Planning Discourse by Modelling Audience Interpretation Strategies
Date: 1992
Abstract:One of the major problems facing designers of computer systems which are to use natural language to achieve their aims is that of finding a way to integrate all of the information which is needed to produce reasonable texts. Amongst these types of information, one of the most complex and hardest to use is knowledge of the intended audience of the text. This thesis proposes an architecture for text planning which places the text planning system's model of its intended audience at the heart of the planning system. The proposed architecture is based upon ideas from research into planning in Artificial Intelligence which are extended to cover some simple types of multi-agent planning. This planning model and the representation which supports it are first discussed informally and then presented in a fairly formal manner. Various simple types of audience model are then encoded within the model and discussed in terms of their effects on the linguistic competence of the resulting text planner. A computer system named Riple has been implemented which embodies the mechanisms discussed in this thesis.

[Search These Pages] [DAI Home Page] [Comment]