MOSAIC: Orchestrating Collaborative Knowledge Tracing with Hierarchical Semantic Alignment

2026-06-27

AI Summary

This work addresses the limitations of traditional knowledge tracing approaches, which rely on shallow ID-based representations and struggle to capture hierarchical knowledge structures and collaborative learning signals. The authors propose MOSAIC, a novel framework that leverages a frozen large language model to generate context-aware, dynamic semantic embeddings. MOSAIC jointly estimates mastery across multiple granularities—concepts, topic clusters, and global proficiency—through hierarchical prediction prompts. It further introduces a cross-granularity consistency objective to explicitly model interdependencies among knowledge levels and peer interactions. Evaluated on ASSISTments, EdNet, and a large-scale MOOC dataset, MOSAIC achieves state-of-the-art performance, with up to 3.4% higher AUC and 2.5% improved accuracy, demonstrating particularly strong gains in long-sequence and high-collaboration settings (e.g., AUC of 0.862 on MOOC data).

Abstract

Knowledge Tracing (KT) is important for personalized education but traditionally suffers from two key limitations: a reliance on shallow ID-based representations that neglect semantic depth and a restriction to single-granularity mastery estimation that overlooks hierarchical knowledge dependencies. To address these challenges, we propose MOSAIC (Multi-granularity Online Semantic AI for Collaborative Knowledge), a novel framework that orchestrates LLM-driven semantic alignment with sequential modeling. Unlike methods that use LLMs solely as predictors, MOSAIC leverages a frozen LLM to generate dynamic, context-aware embeddings and hierarchical prediction prompts, explicitly capturing collaborative signals and peer interactions. Furthermore, we introduce a cross-granularity consistency objective that jointly regularizes mastery estimation across concept, topic-cluster, and global proficiency levels. Extensive experiments on ASSISTments, EdNet, and a newly collected large-scale MOOC dataset demonstrate that MOSAIC establishes new state-of-the-art results. Specifically, our method achieves AUC improvements of up to 3.4\% and Accuracy gains of up to 2.5 \% across all benchmarks. Notably, MOSAIC exhibits superior robustness in collaboration-rich environments and long-sequence scenarios (AUC 0.862 on MOOC), offering both high predictive precision and semantically grounded interpretability.

Problem

• Knowledge Tracing
• Semantic Representation
• Hierarchical Knowledge
• Mastery Estimation
• Personalized Education

Innovation

• Knowledge Tracing
• Large Language Models
• Hierarchical Semantic Alignment
• Multi-granularity Modeling
• Collaborative Learning