mlsanityData to 7 Oct 2026

SCOReD: Student-Aware CoT Optimization for Recommendation Distillation

arXiv 2607.05734 ↗First posted 7 Jul 2026cs.IR cs.AIcited by 0

On the map: LLM reasoning → LLM reasoning for recommendation

Abstract (arXiv)

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%.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in LLM reasoning for recommendation, one of 17 directions in LLM reasoning. Builds on 23 of its 25 references that are in the map. First posted on arXiv 7 Jul 2026.

Builds on

23 of 25 references.

Closest ancestors · 2

Same direction, newest first

  1. 2026GR2: Generative Reasoning Re-ranker8 Feb
  2. 2025OneRec-Think: In-Text Reasoning for Generative Recommendation13 Oct

Other references · 21

  1. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
  2. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  3. 2025DAPO: An Open-Source LLM Reinforcement Learning System at Scalecited by 2,678
  4. 2023Distilling Step-by-Step! Outperforming Larger Language Models with Less Training Data and Smaller Model Sizescited by 710
  5. 2022Recommendation as Language Processing (RLP): A Unified Pretrain, Personalized Prompt & Predict Paradigm (P5)cited by 396
  6. 2026Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Modelscited by 386
  7. 2023Large Language Models are Zero-Shot Rankers for Recommender Systemscited by 307
  8. 2022Teaching Small Language Models to Reasoncited by 279
  9. 2023LLMLingua: Compressing Prompts for Accelerated Inference of Large Language Modelscited by 279
  10. 2025TokenSkip: Controllable Chain-of-Thought Compression in LLMscited by 246
  11. 2024Bridging Language and Items for Retrieval and Recommendation: Benchmarking LLMs as Semantic Encoderscited by 213
  12. 2023Recommender Systems with Generative Retrievalcited by 191
9 more
  1. 2025OneRec: Unifying Retrieve and Rank with Generative Recommender and Iterative Preference Alignmentcited by 167
  2. 2022M6-Rec: Generative Pretrained Language Models are Open-Ended Recommender Systemscited by 152
  3. 2023Symbolic Chain-of-Thought Distillation: Small Models Can Also "Think" Step-by-Stepcited by 144
  4. 2024C3oT: Generating Shorter Chain-of-Thought without Compromising Effectivenesscited by 117
  5. 2026Reasoning Models Struggle to Control their Chains of Thoughtcited by 22
  6. 2025Mitigating Forgetting in LLM Fine-Tuning via Low-Perplexity Token Learningcited by 12
  7. 2026Reasoning Compression with Mixed-Policy Distillationcited by 2
  8. 2026Compress-Distill: Reasoning Trace Compression for Efficient Knowledge Distillationcited by 2
  9. 2026CRISP: Compressing Redundancy in Chain-of-Thought via Intrinsic Saliency Pruningcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024Can Small Language Models be Good Reasoners for Sequential Recommendation?
  2. 2025TrackRec: Iterative Alternating Feedback with Chain-of-Thought via Preference Alignment for Recommendation
  3. 2026SelfDR: Self-Distillation from Reasoning for LLM-Based Recommendation
  4. 2025Towards Efficient CoT Distillation: Self-Guided Rationale Selector for Better Performance with Fewer Rationales
  5. 2025Merge-of-Thought Distillation
  6. 2026WhisperRec: Latent Reasoning for Efficient Foundation Recommendation Models
  7. 2025Reinforced Latent Reasoning for LLM-based Recommendation
  8. 2026rEDMRec: Distilling Large Language Model Reasoning into an Editable Experience Memory for Recommendation