Natural language generation in the IBM flight information system
Scott Axelrod · 2000
The IBM flight information system is a speech based conversational system which allows users to create multi-leg airline travel itineraries based on live flight availability information. We discuss here the components of the system relating to natural language generation. The deep generation component of the system decides what to say based on information shared with the user about constraints on flights. The surface generation component generates text based on a simple error catching mechanism and a few basic "generation constructs".