English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Recommendation as Instruction Following: An LLM-Empowered Recommendation Approach

Forum topic · 小凯 · 2026-07-05

Summary

This May 2023 arXiv paper (arXiv:2305.07001) by Junjie Zhang, Ruobing Xie, Yupeng Hou, Wayne Xin Zhao, Leyu Lin, and Ji-Rong Wen reframes recommendation as an instruction-following task for large language models (LLMs). The authors propose converting user behavior, item information, and recommendation tasks into natural language instructions, enabling a single LLM to handle diverse recommendation scenarios such as rating prediction, sequential recommendation, and zero-shot generalization to unseen tasks and domains. By tuning the LLM on instruction-formatted recommendation data, the approach leverages the semantic knowledge and generalization ability of pretrained language models, addressing limitations of conventional collaborative filtering that struggles with sparse signals, cold start, and task-specific training. The work is an early and influential contribution to the LLM-for-recommendation (Gen-Rec) research direction, showing that instruction tuning can align general-purpose LLMs with recommendation objectives. This article presents the paper's motivation, method design, experimental setup, key findings, and implications for building LLM-based recommender systems, along with related literature on foundation models for personalized ranking.

Recommendation as Instruction Following: An LLM-Empowered Recommendation Approach (arXiv, May 2023)

Paper: https://arxiv.org/abs/2305.07001

Authors: Junjie Zhang, Ruobing Xie, Yupeng Hou, Wayne Xin Zhao, Leyu Lin, Ji-Rong Wen (Renmin University of China / Kuaishou collaboration)

Venue: arXiv preprint, May 2023

Key points

  • Reframes recommendation as an instruction-following problem: user interactions, item metadata, and recommendation tasks are verbalized into natural language instructions that a large language model can directly consume and act upon.
  • Proposes a single LLM-based framework that can be tuned on instruction-formatted recommendation data, unifying multiple recommendation tasks (e.g., rating prediction, sequential/item recommendation) under one model instead of task-specific pipelines.
  • Emphasizes zero-shot generalization: after instruction tuning, the model can handle unseen recommendation tasks and transfer across domains, mitigating cold-start and data-sparsity issues common in collaborative filtering.
  • Leverages the semantic prior of pretrained LLMs (world knowledge, language understanding of item text) to complement or replace traditional ID-based embedding approaches.
  • An early, influential entry in the LLM-for-recommendation line of work, alongside later efforts such as decoder-only foundation models for personalized ranking and language-item retrieval bridging.
  • Motivation

    Traditional recommenders (matrix factorization, deep CTR models, sequential Transformers) learn from user–item interaction IDs. They struggle with sparse feedback, cold-start items and users, and require separate models per task. Meanwhile, LLMs exhibit strong generalization from instruction tuning in NLP. This paper asks: can recommendation itself be cast as instruction following, so that one LLM serves many recommendation scenarios?

    Approach outline

    1. Task verbalization — encode user history, candidate items, and the target task as natural language instructions/templates. 2. Instruction tuning — fine-tune an LLM on instruction–response pairs built from recommendation datasets so the model learns to produce recommendation outputs (e.g., predicted ratings or next items) as text. 3. Inference — at recommendation time, format the user's context as an instruction and let the tuned LLM generate the answer, enabling zero-shot handling of new tasks and domains.

    Evaluation

    The paper evaluates on public recommendation benchmarks, comparing against traditional collaborative filtering and sequential baselines, and specifically tests zero-shot/transfer settings where the model faces tasks or domains not seen during tuning. Readers should consult the original PDF tables for exact quantitative results.

    Takeaways for practitioners

  • Architecture: instruction-tuned LLMs can act as unified recommenders, but latency and inference cost remain hard constraints for production deployment.
  • Data: the quality of instruction-formatted recommendation data is as important as model scale; watch for distribution shift between training tasks and target scenarios.
  • Evaluation: text-generation metrics must be paired with ranking metrics (Recall@k, nDCG) and online signals.
  • Open problems: hallucinated or non-existent item outputs, catalog alignment, and multilingual/multimodal extension.
  • Related entries

  • 360Brew: A Decoder-only Foundation Model for Personalized Ranking
  • Bridging Language and Items for Retrieval and Recommendation (Mar 2024)
  • Data-efficient Fine-tuning for LLM-based Recommendation, SIGIR 2024
Original link: arXiv:2305.07001

Tags

#large-language-models#recommender-systems#instruction-tuning#zero-shot-learning#generative-recommendation#arxiv

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178208924