Advancing Knowledge Tracing by Exploring Follow-up Performance Trends

2025-08-11

AI Summary

Existing knowledge tracing (KT) methods often suffer from correlation conflicts between historical interaction sequences and future performance, degrading prediction accuracy. To address this, we propose a novel KT paradigm that explicitly incorporates students’ future learning trends—marking the first effort to integrate forward-looking trend modeling into KT frameworks. Our approach comprises three key components: (i) construction of forward-looking learning sequences, (ii) linear time-trend retrieval over future trajectories, and (iii) a similarity-aware attention mechanism that jointly leverages frequency-based and context-aware similarity to align historical behaviors with predicted trends. Evaluated on six real-world educational datasets, our method consistently outperforms ten state-of-the-art baselines, achieving an average accuracy of 84.85%—an absolute improvement of 8.74 percentage points. It effectively mitigates correlation conflicts, enhances interpretability of long-term learning progression, and improves prediction robustness.

Technology Category

• Data Mining & Knowledge Management: Knowledge Acquisition from the Web
• Machine Learning: Learning Preferences or Rankings
• Knowledge Representation and Reasoning: Knowledge Acquisition

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Intelligent Tutoring Systems (ITS), such as Massive Open Online Courses, offer new opportunities for human learning. At the core of such systems, knowledge tracing (KT) predicts students' future performance by analyzing their historical learning activities, enabling an accurate evaluation of students' knowledge states over time. We show that existing KT methods often encounter correlation conflicts when analyzing the relationships between historical learning sequences and future performance. To address such conflicts, we propose to extract so-called Follow-up Performance Trends (FPTs) from historical ITS data and to incorporate them into KT. We propose a method called Forward-Looking Knowledge Tracing (FINER) that combines historical learning sequences with FPTs to enhance student performance prediction accuracy. FINER constructs learning patterns that facilitate the retrieval of FPTs from historical ITS data in linear time; FINER includes a novel similarity-aware attention mechanism that aggregates FPTs based on both frequency and contextual similarity; and FINER offers means of combining FPTs and historical learning sequences to enable more accurate prediction of student future performance. Experiments on six real-world datasets show that FINER can outperform ten state-of-the-art KT methods, increasing accuracy by 8.74% to 84.85%.

Problem

• Resolve correlation conflicts in knowledge tracing methods
• Enhance student performance prediction using Follow-up Performance Trends
• Improve accuracy of Intelligent Tutoring Systems' evaluations

Innovation

• Extracts Follow-up Performance Trends from data
• Uses similarity-aware attention mechanism
• Combines trends with learning sequences