國家教育研究院

　　本計畫是一年期計畫，透過模擬研究探討結合半監督式演算法則之電腦化認知診斷測驗演算法之成效，並探討結合選擇題型與建構反應題型之電腦化認知診斷測驗估計成效，最後以實徵資料探討半監督式演算法則之電腦化認知診斷測驗演算法之成效。研究結果顯示:模擬資料和實徵資料均顯示監督式、半監督式與非監督式電腦化認知診斷的估計成效受到訓練樣本數、概念數與試題參數的影響，訓練樣本數愈多，表示加入的訊息量愈多，愈能降低估計的不確定性而提升估計的精準度。當融入建構反應題型估計時，參測猜數限制為0，能得到較佳的估計效果。

　　This study was proposed for one-year period of research implementation. The project focused on cognitive diagnostic models estimation with unsupervised, supervised and semi-supervised learning. With combining multiple choice items and constructed response items in a test, the performance in estimating attributes also explored in this study. A simulation study was conducted to evaluate the performance of the proposed algorithms. A real data was used as an example. The results showed that the accuracies in unsupervised, supervised and semi-supervised learning based on cognitive diagnostic models were affected by training sample size, the number of attributes, and item parameters. The more training sample size has, the higher estimation accuracy was. While adding constructed items in a test, the better performance in estimation attributes under constraining the guessing parameters into zero conditions.