A Training-Free Large Reasoning Model-based Knowledge Tracing Framework for Unified Prediction and Prescription

TL;DR

This paper introduces Thinking-KT, a training-free framework using Test-Time Scaling that enables small LLMs to perform accurate knowledge tracing, feedback, and recommendations in a unified, resource-efficient manner.

Contribution

The paper proposes a novel training-free KT framework with TTS, allowing small LLMs to perform multiple tasks simultaneously without fine-tuning.

Findings

TTS significantly improves LLM-based KT performance.

Small LLMs can serve as unified intelligent tutoring systems.

Unified prediction and prescription are achievable without model fine-tuning.

Abstract

Knowledge Tracing (KT) aims to estimate a learner's evolving mastery based on interaction histories. Recent studies have explored Large Language Models (LLMs) for KT via autoregressive nature, but such approaches typically require fine-tuning and exhibit unstable or near-random performance. Moreover, prior KT systems primarily focus on prediction and rely on multi-stage pipelines for feedback and recommendation, resulting in increased system complexity and resources. To address this gap, we propose Thinking-KT, a training-free KT framework that incorporates Test-Time Scaling (TTS), enabling even small LLMs to achieve competitive KT performance. Moreover, in this framework, a small LLM can jointly perform KT prediction, personalized feedback generation, and learning recommendation in a unified output without degrading prediction accuracy. Beyond performance, we present the systematic analysis of reasoning traces in KT. Our results demonstrate that TTS is a critical yet underexplored factor in LLM-based KT, and that small LLMs can serve as unified ITS engines.