| 06 |
Here, a moderately challenging problem for you!VAE-based individually optimized problem recommendation 
|
|---|
We are researching technologies that recommend to students the optimal learning materials among many options to support individualized learning. In this presentation, we describe our novel method that predicts the probability of correctly answering a first-time problem based solely on the correct/incorrect/unanswered data of many problems by students. This technique, which is applicable regardless of the subject or type of problem, is useful for recommending moderately difficult problems based on the student's ability. We seek a future in which we can provide efficient learning for people of any learning level through technology that recommends optimal individualized learning materials.
[1] T. Hattori, H. Sawada, S. Fujita, T. Kobayashi, K. Kamei, F. Naya, “Monotonic variational autoencoder based individually optimized problem recommender system,” in Comp. Proc. 13th International Learning Analytics and Knowledge Conference (LAK23),2023.
Takashi Hattori
Child Development Research Group, Innovative Communication Laboratory