EVALUATING DIALOGUE STRATEGIES UNDER COMMUNICATION ERRORS USING COMPUTER-TO-COMPUTER SIMULATION

In this paper, experimental results of evaluating dialogue strategies of confirmation with a noisy channel are presented. First, the types o f errors in task-oriented dialogues are investigated and classified as communication, dialogue, knowledge, problem solving, or objective err ors. Since the errors are of different levels, the methods for recover ing from errors must be examined separately. We have investigated that the dialogue and knowledge errors generated by communication errors c an be recovered through system confirmation with the user. In addition , we examined that the manner in which a system initiates dialogue, na mely, dialogue strategies, might influence the cooperativity of their interactions depending on the frequency of confirmations and the amoun t of information conveyed. Furthermore, the choice of dialogue strateg ies will be influenced by the rate of occurrence of communication erro rs in a communication channel and related to the properties of the tas k, for example, the difficulty in achieving a goal or the frequency of the movement of initiatives. To verify these hypotheses, we prepared a testbed task, the Group Scheduling Task, and examined it through a c omputer-to-computer dialogue simulation in which one system took the p art of a scheduling system and the other system acted as a user. In th is simulation, erroneous input for the scheduling system was also deve loped. The user system was designed to act randomly so that it could s imulate a real human user, while the scheduling system was devised to strictly follow a particular dialogue strategy of confirmation. The ex perimental results showed that a certain amount of confirmation was re quired to overcome errors when the rate of occurrence of communication errors was high, but that excessive confirmation did not serve to res olve errors, depending on the task involved.