(BR), 19 CPqD (BR).

Abstract. This paper discussed the implementation of behavior tree technology in behavioral modeling domain. Existing framework can't provide the ability of reasoning while take into account the ability of learning. To solve this problem, we propose a reinforcement learning behavior tree framework based on reinforcement theory. Following our study, a QBot model is build based on the framework in the Raven platform, a popular test bed for game AI development. This paper carried out simulation experiments which include 3 opponent agents. The result shows that QBot outperforms the other 2 Raven_Bots which adopt the default agent model in Raven platform, and thus the result proves that the framework is advanced.

No abstract

Evaluating the quality of a dialogue system is an understudied problem. The recent evolution of evaluation method motivated this survey, in which an explicit and comprehensive analysis of the existing methods is sought. We are first to divide the evaluation methods into three classes, i.e., automatic evaluation, human-involved evaluation and user simulator based evaluation. Then, each class is covered with main features and the related evaluation metrics. The existence of benchmarks, suitable for the evaluation of dialogue techniques are also discussed in detail. Finally, some open issues are pointed out to bring the evaluation method into a new frontier.

No abstract