Case-based retrieval on question items generation

• 第一著者: Subroto, Imam Much
• フォーマット: 学位論文
• 言語: 英語
• 出版事項: 2007
• 主題: QA75 Electronic computers. Computer science
• オンライン･アクセス: http://eprints.utm.my/6654/1/ImamMuchIbnuSubrotoMFSKSM2007.pdf

In the education, the purpose of the conducting test is to determine whether the instructional objectives have been achieved or not. It is a challenge to build a learning system that meet pedagogical aspect of learning. Test items should match to the learning outcomes and the conditions determine by the instructional objectives. Taxonomy Bloom’s known is as the standard of the instructional objective level on the cognitive domain. This work is to study how effective the Case-based Reasoning (CBR) method is to solve the generation of question items problem. CBR is the artificial intelligent method that is suitable to solve the problem by finding similar cases from the past. Based on the similar case, the solution is to reuse the similar case and to revise its similar case solution. It is the fact that some question items or some test may be reused or revised for future situation. This work has been successfully implementing the CBR method on question items generation. Some retrieval techniques (Rule Base Reasoning and CBR) and similarity measure (Nearest neighborhood and Euclidean distance) has been experimented. From these experiments is that, CBR retrieval technique using Euclidean distance similarity and inductive indexing approach is the best performance. The experiment has given the similarity tolerance 0.7 is acceptable because it categorizes to high similarity and the recall is enough to give suggestion solution (in this experiment about 3 or 4 similar cases). Finally the overall results show that the complete task of CBR method has successfully solved the problem of matching the learning outcomes with the instructional objectives.