SCOReD: Student-Aware CoT Optimization for Recommendation Distillation

2026-07-06

AI Summary

This work addresses the limitations of chain-of-thought (CoT) reasoning trajectories generated by teacher models—namely their high uncertainty, redundancy, and distributional mismatch with student models—which hinder the effectiveness of supervised fine-tuning in recommendation systems. To overcome these issues, the paper introduces a novel CoT optimization framework featuring a student-aware mechanism. Specifically, it decomposes teacher-generated reasoning traces into typed segments and dynamically applies strategies such as retention, rewriting, fusion, or pruning based on student attention weights and log-probability gains, thereby reconstructing concise training data aligned with the student’s distribution. Experimental results demonstrate that, compared to baseline supervised fine-tuning methods, the proposed approach improves NDCG and Recall@5 by 1.56% and 1.9%, respectively, while reducing reasoning length by 27.3%.

Abstract

Chain-of-thought (CoT) distillation in the recommendation domain is a necessary precursor to RL training, but raw teacher traces are ill-suited to this task. Large teachers approach the recommendation task with unusually high reasoning uncertainty, repeatedly rechecking their answers without revising them; supervised fine-tuning on such traces produces verbose students that never revise their initial guess. Furthermore, due to the novelty of the recommendation domain, the teacher's reasoning traces are highly out-of-distribution for the small student LLM. We propose Student-Aware CoT Optimization for Recommendation Distillation (SCOReD), a CoT optimization framework tailored to recommendation that first parses each teacher trace into typed segments and uses the student LLM's attention to score the importance of each segment. Then SCOReD dynamically selects a per-segment edit (KEEP / REWRITE / FUSE / PRUNE) based on the output length and comparative log probability lift of the answer given the edit as per the student. Therefore, SCOReD prunes redundant sections of the reasoning trace while preserving information-dense sections and adapts raw teacher traces to the student's output distribution. Training on SCOReD-optimized CoTs provides a cleaner learning signal to the student model and improves over baseline SFT by 1.56% NDCG and 1.9% Recall@5, while reducing reasoning length by 27.3%.

Problem

• Chain-of-Thought Distillation
• Recommendation Systems
• Reasoning Uncertainty
• Out-of-Distribution
• Student-Teacher Learning

Innovation

• Chain-of-Thought Distillation
• Student-Aware Optimization
• Recommendation Systems
• Reasoning Trace Editing
• Large Language Models