KT-XL: A Knowledge Tracing Model for Predicting Learning Performance Based on Transformer-XL.

With the development of artificial intelligence (AI) technology, online teaching systems have become intellectualized. Knowledge tracing is an important task of intelligent teaching system. The goal of knowledge tracing is to trace students' knowledge status based on their past performance. It is widely used for predicting students' performance and building knowledge graph, which plays an important role in constructing adaptive (personalized) teaching systems. Previous models can not deal with subjective problems, and the performance is somewhat poor when input exercise sequences are long. In this paper, we propose a knowledge tracing model for subjective problems based on Transformer-XL, which can predict students' performance of a specific exercise. Specifically, we introduce a recurrence mechanism to the model to capture longer-term dependency and achieve better performance on both short and long sequences. Experiments on multiple real-world data sets and synthetic data sets show that our model performs better in predicting students' performance than state of the art models.