Pith. sign in

REVIEW 22 cited by

A Survey on Large Language Models for Recommendation

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2305.19860 v5 pith:BX2OEESE submitted 2023-05-31 cs.IR cs.AI

Likang Wu , Zhi Zheng , Zhaopeng Qiu , Hao Wang , Hongchao Gu , Tingjia Shen , Chuan Qin , Chen Zhu
show 4 more authors
Hengshu Zhu Qi Liu Hui Xiong Enhong Chen
This is my paper
classification cs.IRcs.AI
keywords recommendationmodelslanguagesystemsexistinggithublargelearning
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Large Language Models (LLMs) have emerged as powerful tools in the field of Natural Language Processing (NLP) and have recently gained significant attention in the domain of Recommendation Systems (RS). These models, trained on massive amounts of data using self-supervised learning, have demonstrated remarkable success in learning universal representations and have the potential to enhance various aspects of recommendation systems by some effective transfer techniques such as fine-tuning and prompt tuning, and so on. The crucial aspect of harnessing the power of language models in enhancing recommendation quality is the utilization of their high-quality representations of textual features and their extensive coverage of external knowledge to establish correlations between items and users. To provide a comprehensive understanding of the existing LLM-based recommendation systems, this survey presents a taxonomy that categorizes these models into two major paradigms, respectively Discriminative LLM for Recommendation (DLLM4Rec) and Generative LLM for Recommendation (GLLM4Rec), with the latter being systematically sorted out for the first time. Furthermore, we systematically review and analyze existing LLM-based recommendation systems within each paradigm, providing insights into their methodologies, techniques, and performance. Additionally, we identify key challenges and several valuable findings to provide researchers and practitioners with inspiration. We have also created a GitHub repository to index relevant papers on LLMs for recommendation, https://github.com/WLiK/LLM4Rec.

Discussion (0). Sign in to comment.

Forward citations

Cited by 22 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Bridging the Information Gap: Semantic Densification and Hindsight Distillation for Cold-Start Prediction

    cs.AI 2026-07 conditional novelty 6.0 of 10

    SemRaD improves cold-start LTV and CVR prediction by encoding LLM-written semantic profiles with a learned gate and distilling a hindsight-reconciled future target through per-user distillation experts, confirmed by a...

  2. ShopX: A Foundation Model for Intent-to-Item Fulfillment in Agentic Shopping

    cs.IR 2026-06 conditional novelty 6.0 of 10

    A single LLM trained to emit semantic item codes can fulfill complex shopping intents with fewer tool hand-offs, improving multi-turn follow-up on Taobao-derived tasks.

  3. SafeGEO: Understanding Generative Engine Optimization Risks in Recommendation Agents

    cs.IR 2026-06 unverdicted novelty 6.0 of 10

    SafeGEO benchmark demonstrates that GEO attacks raise flawed product inclusion in recommendation sets by up to 83.2%, with partial mitigation from defensive prompting and evidence checks.

  4. Modular Representation Compression: Adapting LLMs for Efficient and Effective Recommendations

    cs.IR 2026-04 unverdicted novelty 6.0 of 10

    LLMs exhibit mid-layer representation advantage for recommendations; MARC compresses representations modularly to reduce costs while improving performance, as shown in a large-scale online advertising deployment.

  5. MATRAG: Multi-Agent Transparent Retrieval-Augmented Generation for Explainable Recommendations

    cs.IR 2026-02 unverdicted novelty 6.0 of 10

    MATRAG deploys four agents (user modeling, item analysis, reasoning, explanation) plus knowledge-graph retrieval and a transparency score to raise hit rate 12.7% and NDCG 15.3% while producing explanations rated helpf...

  6. UniRec: Unified Multimodal Encoding for LLM-Based Recommendations

    cs.IR 2026-01 unverdicted novelty 6.0 of 10

    UniRec unifies heterogeneous recommendation modalities via specialized encoders, triplet representations, and hierarchical modeling to outperform prior multimodal LLM recommenders by up to 15% on benchmarks.

  7. Are LLMs Reliable Rankers? Rank Manipulation via Two-Stage Token Optimization

    cs.CL 2025-10 conditional novelty 6.0 of 10

    RAF, a two-stage token-optimization attack, creates brief natural-sounding text injections that reliably boost a target item's rank in LLM reranking outputs, beating state-of-the-art baselines in effectiveness, stealt...

  8. Understanding Generative Recommendation with Semantic IDs from a Model-scaling View

    cs.AI 2025-09 conditional novelty 6.0 of 10

    Semantic-ID-based generative recommenders saturate as model size grows, while directly using an LLM as the recommender keeps improving with scale and learns collaborative filtering signals better.

  9. Beyond Semantic Understanding: Preserving Collaborative Frequency Components in LLM-based Recommendation

    cs.CL 2025-08 conditional novelty 6.0 of 10

    FreLLM4Rec shows that LLMs attenuate low-frequency collaborative components of item embeddings and introduces graph and temporal low-pass filters that preserve them, achieving up to 8% NDCG@10 gains.

  10. LLM-Derived Priors for Thompson Sampling in Cold-Start Comment Recommendation

    cs.IR 2026-08 conditional novelty 5.0 of 10

    LLM-derived Bayesian priors for comment recommendation improve click-through rate in the 10 to 49 impression cold-start range (gender prior +9.5%, content prior +7.8%), but not at the aggregate level.

  11. Tokenizing Numerical and Embedding Features for LLM RecSys

    cs.IR 2026-07 conditional novelty 5.0 of 10

    Interaction-based soft-token fusion of numerical and embedding features beats text-only and direct-concatenation LLM two-tower recommenders on Beauty, Sports and Outdoors, and Toys and Games.

  12. Tokenizing Numerical and Embedding Features for LLM RecSys

    cs.IR 2026-07 conditional novelty 5.0 of 10

    A soft-token fusion framework that turns price and dense-embedding features into LLM-readable tokens improves retrieval in an LLM-based two-tower recommender, and interaction-based fusion beats concatenation on three ...

  13. Tokenizing Numerical and Embedding Features for LLM RecSys

    cs.IR 2026-07 conditional novelty 5.0 of 10

    Interaction-based soft-token fusion of numerical and embedding features improves LLM two-tower retrieval over text-only and direct-concatenation baselines on three Amazon datasets.

  14. ShopX: A Foundation Model for Intent-to-Item Fulfillment in Agentic Shopping

    cs.IR 2026-06 unverdicted novelty 5.0 of 10

    ShopX is a single foundation model combining intent understanding, planning, and SID-native item fulfillment for agentic shopping, with claimed improvements over tool-mediated systems on Taobao logs.

  15. Ocean4Rec: Offline LLM-Derived OCEAN Profiles for Request-Time VOD Reranking

    cs.IR 2026-05 unverdicted novelty 5.0 of 10

    Ocean4Rec uses offline LLM to create OCEAN profiles for items and time-decayed user profiles for request-time numeric reranking, improving NDCG@20 by 7.6% and 61.5% over base+recency in offline VOD evaluations.

  16. A Comparative Study of Demonstration Selection for Practical Large Language Models-based Next POI Prediction

    cs.CL 2026-03 conditional novelty 5.0 of 10

    Heuristic demonstration selection methods outperform embedding-based methods for practical LLM-based next POI prediction on three real-world datasets.

  17. Why Thinking Hurts: Diagnosing and Rectifying Linguistic Inertia in Large Language Models for Recommendation

    cs.IR 2026-02 conditional novelty 5.0 of 10

    Chain-of-thought reasoning degrades semantic-ID recommendation accuracy through 'linguistic inertia,' and a training-free compression-plus-contrastive decoding fix restores and often improves accuracy.

  18. Benchmark Leakage Trap: Can We Trust LLM-based Recommendation?

    cs.LG 2026-02 conditional novelty 5.0 of 10

    Fine-tuning an LLM recommender on a slice of the benchmark inflates AUC/UAUC for in-domain leakage and degrades it for out-of-domain leakage, showing benchmark contamination can distort LLM-based recommendation evaluation.

  19. Diagnosing LLM-based Rerankers in Cold-Start Recommender Systems: Coverage, Exposure and Practical Mitigations

    cs.IR 2026-02 accept novelty 5.0 of 10

    LLM rerankers in cold-start recsys show recall@200 of 0.109, concentrate on only 3 items, and are beaten by popularity baselines (HR@10 0.268 vs 0.008).

  20. TrackRec: Iterative Alternating Feedback with Chain-of-Thought via Preference Alignment for Recommendation

    cs.IR 2025-08 conditional novelty 5.0 of 10

    TrackRec trains a small LLM to generate user-preference summaries and a validator to score them, alternating the training so each improves the other, and reports gains on public and industrial recommendation benchmarks.

  21. Fortress: A Case Study in Stabilizing Search Recommendations via Temporal Data Augmentation and Feature Pruning

    cs.IR 2026-05 unverdicted novelty 4.0 of 10

    Fortress stabilizes query-to-app relevance models by pruning features that cause inconsistent predictions across time periods while retaining predictive power from engagement signals.

  22. A Survey of Personalized Federated Foundation Models for Privacy-Preserving Recommendation

    cs.LG 2025-06 unverdicted novelty 3.0 of 10

    A survey of personalization techniques and foundation model adaptations in federated settings for privacy-preserving recommendations, emphasizing their architectural intersection.

Pith tools