Genetic Algorithm for Bayesian Knowledge Tracing: A Practical Application

Shuai Sun, Xuegang Hu, Chenyang Bu, Fei Liu, Yuhong Zhang, Wenjian Luo

Abstract: Online intelligent tutoring systems have developed rapidly in recent years. Analyzing educational data to help students personalize learning has become a research hotspot. Knowledge Tracing (KT) aims to assess students’ changing cognitive states of skills by analyzing their performance on answers. As a representative KT model, Bayesian Knowledge Tracing (BKT) has good interpretability due to the use of the Hidden Markov Model. However, BKT needs to model students’ performance on different skills separately. If BKT simultaneously traces the cognitive states of students’ multiple skills, its time complexity increases exponentially with the number of skills. Therefore, we introduce a genetic algorithm to solve this problem and propose a Multi-skills BKT. This approach allows the BKT model to handle multiple skills simultaneously. Experiments on real datasets show that the model has a significant improvement in prediction performance over the BKT.