A Study on Intelligent Cloud Diagnostic Test and Adaptive Learning Path Models : Differentiation Rules as an Example

• Main Authors: Liu, Yu-Lung, 劉育隆
• Other Authors: Liu, Hsiang-Chuan
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/38110278590402500542

博士 === 亞洲大學 === 資訊工程學系碩士班 === 100 === This study aims to build up a "Intelligent Cloud Diagnostic Test and Adaptive Learning System (ICDTALS), using the polytomous item structure as the selection strategy to establish a cognitive diagnostic model based on Bayesian network. This system has been applied to the “Differentiation Rules” unit in freshman’s calculus class in order to assess the effectiveness of learning performance. At present most teaching and assessments of calculus are still paper-based. Most of the item types of computerized tests are limited to multiple choice items, and using the constructed response items could obtain further information of student’s learning performance. Although constructed response item is more time-consuming, it allows students to get timely feedback right after finishing the test. Therefore, this study developed a system which combined multiple choice items with constructed response items to record multiple problem solving processes. The system with automatic mechanism is capable of analyzing error patterns of students’ answers, reporting referential results through the use of Bayesian network, and giving remedial instruction. The research design of this study was to give five different treatments to five groups, including four experimental groups and one control group. Each group was given one unique remedial instruction path based on its theoretical framework, and used quasi-experimental design to examine the performance of the treatment. The conclusions were as follows: 1. Bayesian network diagnostic model combining multiple choice items and constructed response items has been approved to be an effective model in terms of predicting student’s concept and error types. Both of the accuracy of the concept and the error type were above 90%. 2. The correct classification rate between the auto-analysis mechanism of constructed response items and the expert judgment achieved nearly 94%, and this result has shown that expert judgment could be replaced by auto-analysis mechanism which having timely feedback to students. 3. The adaptive assessment mechanism based on polytomous item structure could save 20% items under the condition of 95% accuracy. It could efficiently shorten measuring time. 4. When the "Intelligent Cloud Diagnostic Test and Adaptive Learning System (ICDTALS)" used for remedial instruction, the results have shown that the performances of SKILL and BUG structure learning path online remedial instruction were better than SKILL and BUG list learning path online remedial instruction. The SKILL and BUG list learning path online remedial instruction was better than traditional remedial teaching where the whole class was treated as a group.