Learn Before
Concept

Knowledge Query Network for Knowledge Tracing

This paper introduces the Knowledge Query Network (KQN), a knowledge-tracing model motivated by the observation that existing models struggle to represent a student's knowledge state or 'knowledge interaction' — the interaction between the student's knowledge state at time tt and a skill at time t+1t+1. The proposed model (Section 3) is built from a knowledge state query, a knowledge state encoder, and a skill encoder, and is optimized accordingly. The authors also derive a probabilistic skill similarity measure (Section 4), showing that a pairwise distance between two skill vectors relates to the logarithm of their odds ratio. Sections: 1. Introduction 2. Related Work 3. Our Proposed Model 4. Probabilistic Skill Similarity 5. Experiments 6. Results and Analysis 7. Conclusions and Future Work.

0

1

Updated 2026-07-11

Tags

Data Science