研發適性化外語對話學習系統之運作模式－以德語會話系統為例

隨著世界各國極力鼓勵數位學習，陸續出現很多數位教材，面對愈來愈多的數位教材，數位學習面對最重要的問題是數位學習的管理以及數位教材的成效。數位教材最大特色是讓學生建構自己的知識，也就是選擇自己想要課程單元來學習，但這可能也造成數位學習的漏洞：由於學生自己可以選擇單元來學習，造成學生可能因為個人學習習慣及偏好，導致每次只有學習某些單元，而無法掌握所學習的數位教材之整體性及補強自己的學習弱點。本論文設計針對數位學習全方位的方式，利用記錄學生的學習過程，藉由人工智慧在學生學習過程中掌握學生對教材的反應，辨識學生學習弱點及缺失，而在下次學習時，針對同一學生補強其弱點，藉而讓學生學習更全面化。本論文雖然以德語會話為範例，但是所研發系統模式適合各種語言會話數位教材。

關鍵字

適性化學習；外語對話；數位內容；外語對話設計

並列摘要

Almost all the countries in the world make every effort to encourage digital learning. The biggest advantage of the digital teaching material is that the student is able to choose any unit he wants to learn. However, this possibility also constitutes a weakness of the digital learning. Regarding the content of digital learning, as the student can choose any units he wants to learn, it is possible that the student only learns some units and will not learn the whole of the digital teaching material because of personal learning habits and preferences. So he cannot make good use of a limited period of time to compensate for his own weaknesses. This thesis aims to develop an adaptive digital conversation system which can record the learning process of the student, control student reaction to teaching material by artificial intelligence algorithms, recognize the weakness of the student’s learning behaviors, and at the next time, target the same student's weaknesses and help him improve by giving him other materials at the same level. With the adaptive learning system the student can achieve better learning results. Although this thesis takes German conversation as an example, the model of the learning system can be developed with various languages.