隨著資訊科技進步及網路發展，人們開始利用數位學習平台提升知識。知識追蹤是數位學習系統的核心，根據學習者的學習紀錄追蹤知識掌握程度並預測學習成效。但目前知識追蹤模型仍存在模型複雜及未考慮遺忘因素等問題。因此，本研究嘗試使用指數平滑法進行改善並預測學習成效及設計開發一個學習儀表板，讓學習者及教師迅速的透過視覺化分析瞭解學習狀態。 本研究採取問卷調查法透過科技接受模式以探討不同背景的學習者對於本研究所開發之視覺化學習儀表板ES-Dashboard之知覺有用性及知覺易用性差異，透過Suvery Cake網路問卷發放進行資料蒐集，回收有效樣本共計64份，藉由敘述性統計、單因子變異數分析等統計方法進行分析。研究結果顯示ES-Dashboard具備不錯之知覺可用性及易用性，且職業對學習儀表板的「知覺有用性」與「知覺易用性」有顯著影響；開放問題分析結果為學習者認為不僅可以在ES-Dashboard迅速了解當前的學習狀態，更可透過系統視覺化圖表快速掌握學習重點、分配學習時間並進行進度追蹤。 除此之外，模型驗證結果顯示本研究所提出基於指數平滑法的知識追蹤模型(ES)不論在整體或是分群情況下，模型指標RMSE(12.57<19.01)及MAE(11.2<17.61)皆優於傳統知識追蹤模型(DKT)。;Along with the advance in information technology and the popularity of the internet, people have started using e-learning platforms to enhance their knowledge. Knowledge tracing is the core of digital learning systems, tracking learners′ knowledge mastery and predicting learning outcomes based on their learning records. However, current knowledge tracing models face issues such as complexity and neglect of forgetting factors. Therefore, this study attempts to use the exponential smoothing method to improve and predict learning outcomes and design and develop a learning dashboard, allowing learners and teachers to quickly understand learning statuses through visual analysis. This study adopts a questionnaire survey method based on the Technology Acceptance Model to explore the differences in perceived usefulness and perceived ease of use of the visualized learning dashboard, ES-Dashboard, developed in this study among learners with different backgrounds. Data was collected through an online questionnaire distributed via Suvery Cake, with a total of 64 valid samples collected. The data was analyzed using descriptive statistics and one-way ANOVA. The results indicate that the ES-Dashboard has good perceived usability and ease of use, and that occupation significantly influences the perceived usefulness and perceived ease of use of the learning dashboard. Open-ended question analysis revealed that learners believe they can quickly understand their current learning status through the ES-Dashboard, and also quickly grasp key learning points, allocate learning time, and track progress through the system′s visual charts. Furthermore, the model validation results indicate that the knowledge tracing model proposed in this study based on exponential smoothing (ES) outperforms the traditional knowledge tracing model (DKT) in both overall and subgroup scenarios, with RMSE (12.57 < 19.01) and MAE (11.2 < 17.61).
