Exploring the Effectiveness of Two Agents and One Agent Dialogue Based Intelligent Tutoring Systems-Taking Pythagoras Theorem as An Example

• Main Authors: YU, HSIN-JUNG, 游欣容
• Other Authors: KUO, BOR-CHEN
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/b69e2b

碩士 === 國立臺中教育大學 === 教育資訊與測驗統計研究所 === 107 === Nowadays, the way of learning has gradually changed. In order to improve teaching model and enhance learning ability, we take advantage of Conversational-based Intelligent Tutoring system to analyze students’ learning information we gathered from different perspective and integrate it to teachers as a reference. This study used Mathematics Conversational-based Intelligent Tutoring System, which is a platform developed by the research team leading by Professor Bor Chen Kuo (National Taichung University of Education, Graduate Institute of Educational Information and Measurement). The system uses its Single and Dual Agent AutoTutor to work on remedial instruction for Pythagorean Theorem and evaluate the effect. The AutoTutor Agent raises questions regarding the teaching materials, and shows proper learning path on the basis of learner’s response. Then the learner can get guidance and feedback from AutoTutor. Therefore, we can understand the difference of teaching effect between Single and Dual Agent AutoTutor. Furthermore, in affection domain, take study interests and self-regulated learning system into consideration, we designed the questionnaire investigating the acceptance and interaction of students using this AutoTutor system to understand learner’s response and perspective. Participants of the study were eighth graders in Taichung City. They were divided into three groups to go on remedial instruction for Pythagorean Theorem. Group A and B use Single and Dual Agent AutoTutor System separately, and Group C use tradtional class learning model. Same teaching materials were used for the three groups. The instruction hours are three class periods. The results showed that Group A,Group B and Group C were all doing well. However, the effects of AutoTutor System were superior to traditional class teaching. Moreover, by conducting Johnson-Neyman Procedure to analyze Group A and Group B students, we find out that when the pre-test scores of the student were under 47, the effect of students using Dual Agent AutoTutor System is obviously better than that of Single Agent AutoTutor System. Also, Dual Agent AutoTutor can attract study interests of students who got lower achievement. Therefore, using Dual Agent AutoTutor System to go on remedial instruction is effective to help students learn better.