Importing Human Tutor’s Conversation Strategy into Dialog Systems for Language Learning using Fuzzy Logic

Foreign language tutors have some conversation strategies to help learners to successfully finish a learning session. But these strategies are  generally described by vague linguistic terms so that they cannot be easily transformed to a computational model. Therefore, this study presents a fuzzy logic-based method that considers learner’s proficiency level and difficulty of system responses together to determine the best desirable system response. By modeling and utilizing human tutor’s conversation strategy, our system is better able to fulfill its role as a conversational agent for language tutoring.