음성 대화 시스템을 위한 예제 기반 대화 관리 방법

Abstract

A generic dialog management framework has been extensively required to build various spoken dialog systems in the real world. This dissertation proposes an example-based approach for a generic dialog management framework in the spoken dialog systems. The basic idea of this approach is to find the most similar example given the current dialog state using heuristic similarity functions. This approach has been applying to implement multiple applications such as car navigation, weather information, TV program guidance, and chatbot for both information access and entertainment. Moreover, we desgiend and implemented a multi-domain dialog system based on this approach to simultaneously manage task-oriented and chat dialogs. In practice, traditional spoken dialog systems have still difficulty selecting which action to take in a given situation because recognition and understanding errors are prevalent due to noise and unexpected inputs. To solve this problem, this dissertation also presents a hybrid approach to improving the robustness of the example-based dialog model by using agenda-based dialog management techniques. Given the agenda graph and multiple recognition hypotheses, the system can predict the next action to maximize multi-level score functions and trigger error recovery strategies to handle exceptional cases due to misunderstandings or unexpected focus shifts. The proposed methods were tested by developing a spoken dialog system for a building guidance domain in an intelligent robot. This system was then evaluated by simulated and real users. The experimental results show that our methods can effectively develop robust dialog management for spoken dialog systems.