On the limited memory BFGS method for large scale optimization
Mathematical Programming: Series A and B
GroupLens: an open architecture for collaborative filtering of netnews
CSCW '94 Proceedings of the 1994 ACM conference on Computer supported cooperative work
Profiling students' adaptation styles in Web-based learning
Computers & Education
E-Commerce Recommendation Applications
Data Mining and Knowledge Discovery
Logically Optimal Curriculum Sequences for Adaptive Hypermedia Systems
AH '00 Proceedings of the International Conference on Adaptive Hypermedia and Adaptive Web-Based Systems
Curriculum Sequencing in a Web-Based Tutor
ITS '98 Proceedings of the 4th International Conference on Intelligent Tutoring Systems
Decision Support Models for Composing and Navigating through e-Learning Objects
HICSS '03 Proceedings of the 36th Annual Hawaii International Conference on System Sciences (HICSS'03) - Track1 - Volume 1
Personalized e-learning system using Item Response Theory
Computers & Education
IEEE Transactions on Knowledge and Data Engineering
Hi-index | 0.00 |
When students study for multiple-choice cloze tests as the Test of English for International Communication (TOEIC), they tend to repeatedly tackle questions of the same type. In such situations, students can effectively solve questions related to their incorrectly answered questions. However, since they need several different kinds of knowledge and a large vocabulary to derive answers, it is inappropriate to statically define the relations among questions from various viewpoints beforehand. In this paper, we propose a recommendation algorithm for English multiple-choice cloze questions that maximize students' expected improvements of test scores based on the learning log data of other students. Effective questions may be identical for most students who incorrectly answered the same questions. Therefore, in our approach, relations among questions in tests and questions studied during tests are determined based on the change from incorrect to the correct answers of the test questions. Questions that maximize the expected test scores, which are calculated based on the input test scores using regression models, are recommended for future students. Based on this method, students can acquire higher test scores with better learning efficiency. Experimental results show that our method yields major improvements in performance compared with random material recommendation method.