人工智慧的實踐：對話式日語學習智慧家教系統進行日語補救教學之研究

Artificial Intelligence in Practice: Conversation-Based Intelligent Tutoring System Conducts Research on Japanese Remedial Teaching

本研究的主要目的是為日語學習者建立以對話為主的智慧家教系統CITS-JL，以期提升臺灣學生在日語學習中的學習效果。本研究以JF Standard日語學習標準為基礎，提出了初級日語學習的知識結構地圖，根據此知識地圖，臺灣學生可以按照22個學習場景以及86個設計的學習單元逐步學習日語。在這項研究中，CITS-JL首先診斷學生的錯誤模式，然後導入適當的認知失衡教學策略來設計對應的問題和對話，藉以釐清學生在學習上可能存在的語詞誤用。此外，CITS-JL考慮了學習系統、研究目標、研究現況這3個向度內容，讓智慧家教系統發展出更豐富的層次，也能夠更完整的協助學生學習與老師教學。根據研究結果表示，CITS-JL可以立即釐清學生的語詞誤用、培養自學的習慣、藉由反覆練習以提高學習效率。此外，此系統可以應用於個人化學習，學生可以在教室外隨時隨地學習日語，可以節省更多的時間和金錢。本研究的貢獻，不僅是為日語學習者建立一個學習工具，而且為教師提供了適當的指導方法。

The objective of this study is to construct a conversation-based intelligent tutoring system for Japanese learning (CITS-JL), and it is expected to improve students’ Japanese learning effects. Based on the Japanese learning standards in the Japanese Foundation Standard, this study proposes a knowledge map for primary Japanese learning; students can learn Japanese step-by-step by following 22 learning scenarios and 86 designed learning units. CITS-JL first diagnosed students’ error patterns and then used appropriate cognitive conflict teaching strategies to design corresponding problems and conversations to clarify possible misuse of words in learning. In addition, CITS-JL considers the learning system, research objectives, research status of three orientation content, so that the intelligent tutoring system develop a richer level, but also to more complete to assist students to learn. The results show that CITS-JL can immediately clarify the misuse of students’ words, cultivate self-study habits, and improve learning efficiency through repeated exercises. In addition, CITS-JL can be applied to personalized learning. Students can learn Japanese anywhere outside the classroom, saving more time and money. This study contributes to research by building a learning tool for Japanese learners, providing teachers with appropriate guidelines for remedial teaching.