Dialogues from Texts: How to Generate Answers from a Discourse Model

In this paper, we shall deal with a system for text generation and understanding called GETA_RUN (GEneration of Text and Analysis for Reference Understanding). Dialogues in our system may be regarded as a sequence of question-answering actions with the aim to retrieve as much information as possible from a text. In a certain sense the system acts as a very intelligent information retrieval system which allows to recover the contents of a text on the basis of what is actually being linguistically expressed. No implicit information is recoverable from the system, however. The system works by encoding the linguistic contents of a text fully analyzed in the Discourse Model where all entities, properties and relations are recorded. Further semantic properties and relations are then transferred into a Knowledge Representation Module running under KL-One to be used for consistency check whenever the dialogue is started up.