Nonverbal action selection for explanations using an enhanced behavior net


In this paper we present a novel approach to the nonverbal action selection problem for an agent in an intelligent tutoring system. We use a variation of the original Maes' Behavior Net that has several improvements that allow modeling action selection using the content of the utterance, communicative goals, and the discourse history. This Enhanced Behavior Net can perform action selection dynamically, reprioritize actions based on all these elements, and resolve conflict situations without the use of sophisticated predefined rules. © 2011 Springer-Verlag.

Publication Title

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)