| 摘要: | 數位學習的發展成為一種新型態的學習模式，加上新型冠狀病毒的疫情讓各國對數位學習的發展更加重視。在2020年時疫情的大爆發導致各國封城狀態，使得全球15億學生無法到校學習，數位學習的方式就顯得更加重要。根據問卷調查顯示數位學習時，學習者有更多的環境誘惑會使得學習者的學習成效比實體教學時來的更低；因此本研究提出一個「智慧型推薦系統」，是利用Dlib所建立的人臉模型加入學習者的學習表情後進行情緒分析，表情分別為中性、微笑、疑惑及驚訝四種。本研究針對其中兩種在進行課程學習時的表情，給予即時的輔助以提高學習成效，第一種是學習者表現出疑惑等的困惑表情，會擷取畫面並給予該章節額外輔助內容；第二種是在觀看數位教材或是線上測驗學習時，針對內容與錯誤題目的驚訝表情，系統會推薦給予該題型的輔助學習內容。經實驗組與控制組進行實驗，以及相關的問卷顯示：根據本研究實驗組與控制組結果可以發現，有使用推薦系統功能的實驗組在後測的成績上比對照組進步的明顯許多，尤其對後1/3學生有著大幅度的進步，平均分數從41.78分進步到60.29分，平均進度18.51分。本研究之「智慧型推薦學習系統中的額外輔助內容可以幫助我在學習時遇到更少的困難」對「利用智慧型推薦學習系統讓我提高學習時的成效」的相關係數為0.978，顯示是高度相關的，因此可以認定為研究提供的推薦學習系統在學生遇到困難時提供的額外輔助學習內容可以幫助學生更了解該問題所在點並解決問題點。分析結果可以表明推薦功能是可以幫助學習成效的。
The development of digital learning as a new form of learning, coupled with the new coronavirus epidemic, has increased the importance of digital learning in all countries. In the year 2020, the epidemic has led to the closure of countries, making it impossible for 1.5 billion students worldwide to attend school, making digital learning even more im-portant. According to the questionnaire, the learners have more environmental temptations in e-learning, which makes the learners' learning less effective than in physical teaching. Therefore, this study proposes an "intelligent recommendation system" by adding the learners' learning expressions, which are neutral, smiling, puzzled, and surprised, using the face model developed by Dlib for emotion analysis. The first one is the expression of confusion when the learner shows doubt, and the system will capture the picture and give additional content for that chapter. The second one is the expression of surprise when the learner is watching the digital textbook or online test, and the system will recommend the content for that topic. According to the results of the experimental and control groups and the related questionnaires, the experimental group with the recommendation system showed a significant improvement in the post-test scores compared to the control group, especially for the last one-third of the students, with the average score improving from 41.78 to 60.29 with an average progress of 18.51 points. The correlation coefficient be-tween "The additional supplemental content in the intelligent recommendation learning system helps me to have less difficulties in learning" and "Using the intelligent recom-mendation learning system makes me more effective in learning" in this study is 0.978, which shows a high correlation, so it can be concluded that the additional supplemental learning content provided by the recommendation learning system when students en-counter difficulties can help students to better understand The results of the analysis show that the recommendation function is a useful tool to help students understand the problem and solve it. The analysis of the results shows that the recommendation function can help the learning effectiveness. |
