Increasing efficiency of the development of user models

Abstract

This paper introduces Herbal, a high-level behavior representation language for creating AI agents and cognitive models. It describes the lessons from other high-level modeling languages that informed the design of Herbal and that will inform other high-level behavior representation languages. We describe a model built in Herbal to illustrate its use and application. The paper concludes that languages like Herbal can help explain the design intent of intelligent agents and cognitive models, and make them easier to create, modify, and understand. These results appear to be particularly true where the model reuses a lot of its own structures.