Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

GR-LLMs: Recent Advances in Generative Recommendation Based on Large Language Models

T0 review · 3 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read This survey claims that LLM-based generative recommendation is emerging as a new paradigm that can unify recall, ranking, and final list generation in one model, replacing today's cascaded recommender systems.

desk verdict A useful, current survey of industrial LLM-based generative recommendation whose main weakness is treating vendor-reported lifts as established facts; worth refereeing with a modest revision. read the letter →

arxiv 2507.06507 v2 pith:TL2QJ2W4 submitted 2025-07-09 cs.IR cs.AI

Zhen Yang , Haitao Lin , Jiawei xue , Ziji Zhang This is my paper
classification cs.IRcs.AI
keywords generativerecommendationlargelanguagemodelsrecommendersystemsnext-tokenpredictionend-to-endsemanticIDscascadedindustrial
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This survey argues that LLM-based generative recommendation is forming a new paradigm, distinct from the discriminative models and hand-crafted feature pipelines that run most recommender systems today. It organizes recent work into three application settings: generative recall, generative ranking, and end-to-end recommendation that replaces the entire cascade with one model. It also catalogs the training pipelines used, from single-stage next-item prediction to multi-stage alignment with direct preference optimization, along with efficiency and cold-start techniques that make industrial deployment possible. A sympathetic reader would take away that a single sequence-generation model could soon serve as the unified engine for recall, ranking, and final list generation in practical systems.

What carries the argument

The central object is the generative formulation of recommendation as next-token prediction over user behavior sequences; the survey's defining move is to treat item IDs, or semantic IDs, as tokens and the user's interaction history as a sequence, so that recommendation becomes a language-model-style generation problem. This formulation unifies recall and ranking into a single model, and it is what the taxonomy of prompt-based, token-based, embedding-based, hybrid, and end-to-end systems hangs on. The cascade of recall-pre-ranking-ranking serves as the contrasting architecture that generative recommendation is claimed to replace.

What would settle it

Run an end-to-end generative recommender such as OneRec or EGA-V2 on a public benchmark with the same data and compute as a strong cascaded baseline; if it does not match or beat the baseline, the replacement claim is contradicted. Independently, replicating these systems at other platforms would test whether the reported industrial lifts generalize beyond their original settings.

Watch

Extended reading notes

Core claim

The paper's central claim is that the next-token prediction objective that powers LLMs can carry recommendation itself: a user's behavior history becomes a token sequence, items become vocabulary (often semantic IDs produced by quantized autoencoders such as RQ-VAE), and the recommendation problem becomes "predict what comes next." Around that formulation, the survey traces a taxonomy: recall methods that use prompts, tokens, or embeddings; ranking methods that are either fully generative or hybrid integrations with classical CTR models; and end-to-end methods that generate whole recommendation lists, with OneRec and EGA-V2 as the industrial exemplars. It reports that these systems have achieved significant business gains in the past year and argues that scaling laws and world knowledge give generative recommendation a higher ceiling than the traditional cascade.

Load-bearing premise

The survey's central picture depends on the reported industrial gains of the systems it cites being real, reproducible, and general, since no code, data, or independent replication backs those reports.

Editorial extensions

If this is right

  • Recall and ranking can be performed by a single generative model rather than by separate cascaded stages.
  • End-to-end recommendation, trained with preference alignment, can directly output a final list and avoid error propagation across stages.
  • Scaling model size and sequence length should keep improving recommendation quality, something traditional models struggle with.
  • Cold-start items and users can be handled using the world knowledge and multimodal understanding of LLMs.
  • Industrial effort shifts from hand-crafted feature engineering to sequence organization, token design, and post-training alignment.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural extension of this view is that the main design surface of recommender systems becomes item tokenization and data curation rather than feature engineering, so recommender research will start to resemble language-model research.
  • The semantic-ID line of work suggests a testable corollary: better item tokenizers, analogous to tokenizers in NLP, should improve generative recommendation, and future benchmarks should isolate tokenizer quality from model quality.
  • The reported scaling gains come from models up to about one billion parameters, so the obvious next experiment is to train at substantially larger scale and see whether the observed returns continue.
  • The hybrid systems described in the survey suggest the transition may be gradual, with generative models first acting as feature producers for existing cascades before fully replacing them.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper surveys recent work on LLM-based generative recommendation (GR), organizing the field into application settings (recall, ranking, end-to-end), training pipelines (single-stage vs. multi-stage), inference-efficiency techniques, and cold-start/world-knowledge mitigation. It argues that GR is a new paradigm distinct from discriminative recommendation and has strong potential to replace traditional cascaded systems, citing recent industrial systems such as OneRec, EGA-V1/V2, MTGR, and KuaiFormer as evidence. The survey includes a comparative table of training objectives and strategies and concludes with future directions on model scaling, data cleaning, and unified models.

Significance. The paper's main strengths are its timely taxonomy and broad coverage of a rapidly evolving set of 2024-2025 preprints. Table 1 provides a useful comparative summary of training strategies, and the inference-efficiency section collects concrete techniques (e.g., M-FALCON, cluster attention, semantic IDs) that are otherwise scattered across separate papers. If published with appropriate caveats, the survey would be a valuable entry point for researchers entering this area. Its central claim, however, that GR is mature enough to replace cascaded industrial systems rests on uncritically repeated vendor-reported results from internal A/B tests and offline evaluations with no public artifacts or third-party replication. Because this claim is the paper's main thesis, the lack of critical hedging is a substantive issue rather than a cosmetic one.

major comments (3)
  1. [Section 3, opening sentence] The sentence 'In the past year, various GR systems have achieved significant business benefits in practical industrial settings' is stated as established fact, but the evidence consists of industry technical reports (OneRec, EGA-V1/V2, MTGR, KuaiFormer, LEARN) describing internal A/B tests or offline metrics on proprietary data with no public code, data, or third-party replication. This evidence is load-bearing because Sections 3.1-3.3 and 4.1 use these results to support the paper's replacement-of-cascaded-systems thesis. The manuscript should rephrase these statements as reported or vendor-claimed benefits, add an explicit caveat about the lack of independent validation, and distinguish peer-reviewed evidence from preprint/industry reports throughout. The Limitations section only disclaims coverage of recent work, not the reliability of the cited industrial results, so this gap is currently unaddressed.
  2. [Section 3.3] The claim that OneRec 'first deploys the end-to-end recommendation model to the industrial scenario' and 'completely replaces the cascaded online recommendation pipeline with a single model' is presented without reporting the comparison setting, the strength of the baseline, the duration of the online experiment, or the statistical significance of the reported feedback metrics. Given that no external replication is available, the survey should attribute this claim to the OneRec paper and flag it as a single-company A/B result. The same unhedged presentation appears in the descriptions of EGA-V1 and EGA-V2 in Sections 3.3 and 4.1.2, where 'notable improvements' and 'the overall expected return' are reported without qualification.
  3. [Sections 3.2 and 5.1] The paper's treatment of scaling laws is internally in tension. Section 3.2 states that GR 'marks the first observation of the scaling laws inherent in LLMs within large-scale recommendation systems' (citing Zhai et al., 2024), while Section 5.1 admits that 'in these works, the model sizes are still limited to a relatively small level, such as 0.x B or 1.x B, and the performance improvements of models at much larger sizes have not been well validated.' The survey should reconcile these statements: either specify exactly what scaling behavior has been observed at what model sizes, or explicitly weaken the claim that scaling laws are a validated foundation for GR. As written, the earlier assertion overstates the strength of the evidence.
minor comments (6)
  1. [Figure 1] The stage-size annotations appear garbled ('~10!', '~10"', '~10"'), presumably rendering order-of-magnitude labels or exponents; please fix the figure so the intended values are readable.
  2. [Section 2.2] The sentence 'the effectiveness of each isolated stage serves as the upper bound for the subsequent stage' is unclear; the intended meaning appears to be that errors introduced in a stage constrain the performance of all downstream stages, but the current phrasing should be rewritten for precision.
  3. [Section 4.1.2, LUM description] The text 'the third stage trains DLMs for recall and ranking' contains an apparent typo: 'DLMs' should likely be 'DL models' or 'deep learning models.'
  4. [References and body text] There are several formatting inconsistencies: 'RQ-V AE' contains a spurious space in Sections 1 and 4.2, and the Radford et al. reference lists the publisher as 'PmLR' instead of 'PMLR.' Please correct these.
  5. [Section 4.1.2] The heading 'Representation-based finetuning' uses lowercase 'finetuning' while the rest of the paper uses 'fine-tuning'; please unify the terminology.
  6. [Section 3.1, SyNeg] SyNeg is described as a prompt-based recall method, but the original work focuses on generating synthetic hard negatives for dense retrieval; please clarify how this transfers to the recommendation recall stage so that the categorization is accurate.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation; the survey's claims rest on external systems rather than the authors' own fitted results, with only one non-load-bearing self-citation.

full rationale

This paper is a narrative survey: it introduces a taxonomy (GR vs. cascaded systems), organizes recent systems under recall/rank/end-to-end categories, and synthesizes reported industrial results. There are no fitted parameters, no new experiments, and no formal derivation whose conclusion could be equivalent to an input. Its central claim that LLM-based GRs are a distinct paradigm with potential to replace hand-crafted-feature cascades is grounded in external works such as Zhai et al. (2024), Deng et al. (2025), Qiu et al. (2025), Zheng et al. (2025), Han et al. (2025), and Jiang et al. (2025), not in the authors' own prior theorems or fitted values. The only self-citation is Xue et al. (2025), cited in Section 1 among several examples of deep-learning-based recommendation; it is not a premise of any GR conclusion and therefore does not constitute load-bearing self-citation. The Section 3 statement that 'various GR systems have achieved significant business benefits' relies on self-reported industrial evaluations without public replication; that is a verifiability limitation, not circularity, because the survey does not derive those benefits from its own assumptions. The Limitations section likewise only disclaims coverage completeness. No equation, definition, or reported prediction reduces to the survey's own inputs, so no specific circular step can be exhibited.

Assumptions & free parameters 0 free parameters · 0 assumptions · 0 invented entities

The paper introduces no free parameters, no mathematical axioms, and no invented entities. Its claims are descriptive statements about the state of the literature, which depend on trusting the published and arXiv-posted reports of other research groups.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GR-LLMs: Recent Advances in Generative Recommendation Based on Large Language Models." pith.science (2026). https://pith.science/paper/TL2QJ2W4

@misc{pith2026250706507,
  author       = {Pith},
  title        = {Pith review of: GR-LLMs: Recent Advances in Generative Recommendation Based on Large Language Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TL2QJ2W4}},
  note         = {Machine review of arXiv:2507.06507}
}
read the original abstract

In the past year, Generative Recommendations (GRs) have undergone substantial advancements, especially in leveraging the powerful sequence modeling and reasoning capabilities of Large Language Models (LLMs) to enhance overall recommendation performance. LLM-based GRs are forming a new paradigm that is distinctly different from discriminative recommendations, showing strong potential to replace traditional recommendation systems heavily dependent on complex hand-crafted features. In this paper, we provide a comprehensive survey aimed at facilitating further research of LLM-based GRs. Initially, we outline the general preliminaries and application cases of LLM-based GRs. Subsequently, we introduce the main considerations when LLM-based GRs are applied in real industrial scenarios. Finally, we explore promising directions for LLM-based GRs. We hope that this survey contributes to the ongoing advancement of the GR domain.

Figures

Figures reproduced from arXiv: 2507.06507 by the authors.

Figure 1
Figure 1. A typical cascade ranking system, which in [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Exponential Reward Weighting for Fine-Tuning Generative Recommenders under Sparse and Noisy Feedback

    cs.IR 2026-08 conditional novelty 4.0 of 10

    Exponential reward weighting with a tuned temperature improves offline generative recommenders, and a new theory decomposes its suboptimality into coverage and noise costs that predict the observed inverted-U in performance.

Reference graph

Works this paper leans on

92 extracted references · 34 canonical work pages · cited by 1 Pith paper

  1. [1]

    Gediminas Adomavicius and Alexander Tuzhilin. 2005. Toward the next generation of recommender systems: A survey of the state-of-the-art and possible extensions. IEEE Transactions on Knowledge and Data Engineering, 17(6):734--749

    2005

  2. [2]

    Muhammad Arslan, Hussam Ghanem, Saba Munawar, and Christophe Cruz. 2024. A survey on RAG with LLMs . Procedia Computer Science, 246:3781--3790

    2024

  3. [3]

    John S Breese, David Heckerman, and Carl Kadie. 2013. Empirical analysis of predictive algorithms for collaborative filtering. ArXiv preprint arXiv:1301.7363

    work page Pith review arXiv 2013

  4. [4]

    S \'e bastien Bubeck, Varun Chandrasekaran, Ronen Eldan, Johannes Gehrke, Eric Horvitz, Ece Kamar, Peter Lee, Yin Tat Lee, Yuanzhi Li, Scott Lundberg, and 1 others. 2023. Sparks of artificial general intelligence: Early experiments with GPT-4 . ArXiv preprint arXiv:2303.12712

    arXiv 2023

  5. [5]

    Christopher JC Burges. 2010. From RankNet to LambdaRank to LambdaMART : An overview. Learning, 11(23-581):81

    2010

  6. [6]

    Jianxin Chang, Chenbin Zhang, Zhiyi Fu, Xiaoxue Zang, Lin Guan, Jing Lu, Yiqun Hui, Dewei Leng, Yanan Niu, Yang Song, and 1 others. 2023. TWIN : Two-stage interest network for lifelong user behavior modeling in CTR prediction at Kuaishou . In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 3785--3794

    2023

  7. [7]

    Ching-Wei Chen, Paul Lamere, Markus Schedl, and Hamed Zamani. 2018. Recsys challenge 2018: Automatic music playlist continuation. In Proceedings of the 12th ACM Conference on Recommender Systems, pages 527--528

    2018

  8. [8]

    Junyi Chen, Lu Chi, Bingyue Peng, and Zehuan Yuan. 2024 a . HLLM : Enhancing sequential recommendations via hierarchical large language models for item and user modeling. ArXiv preprint arXiv:2409.12740

    arXiv 2024

Show all 92 references
  1. [9]

    Qiwei Chen, Changhua Pei, Shanshan Lv, Chao Li, Junfeng Ge, and Wenwu Ou. 2021. End-to-end user behavior retrieval in click-through rateprediction model. ArXiv preprint arXiv:2108.04468

    2021 arXiv

  2. [10]

    Xu Chen, Yongfeng Zhang, and Zheng Qin. 2019. Dynamic explainable recommendation based on neural attentive models. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 53--60

    2019

  3. [11]

    Yuxin Chen, Junfei Tan, An Zhang, Zhengyi Yang, Leheng Sheng, Enzhi Zhang, Xiang Wang, and Tat-Seng Chua. 2024 b . On softmax direct preference optimization for recommendation. ArXiv preprint arXiv:2406.09215

    2024 arXiv

  4. [12]

    Paul Covington, Jay Adams, and Emre Sargin. 2016. Deep neural networks for YouTube recommendations. In Proceedings of the 10th ACM Conference on Recommender Systems, pages 191--198

    2016

  5. [13]

    Wenqian Cui, Dianzhi Yu, Xiaoqi Jiao, Ziqiao Meng, Guangyan Zhang, Qichao Wang, Yiwen Guo, and Irwin King. 2024. Recent advances in speech language models: A survey. ArXiv preprint arXiv:2410.03751

    2024 arXiv

  6. [14]

    Damai Dai, Chengqi Deng, Chenggang Zhao, RX Xu, Huazuo Gao, Deli Chen, Jiashi Li, Wangding Zeng, Xingkai Yu, Yu Wu, and 1 others. 2024. DeepSeekMoE : Towards ultimate expert specialization in mixture-of-experts language models. ArXiv preprint arXiv:2401.06066

    2024 arXiv

  7. [15]

    Jiaxin Deng, Shiyao Wang, Kuo Cai, Lejian Ren, Qigen Hu, Weifeng Ding, Qiang Luo, and Guorui Zhou. 2025. OneRec : Unifying retrieve and rank with generative recommender and iterative preference alignment. ArXiv preprint arXiv:2502.18965

    2025 arXiv

  8. [16]

    Seungheon Doh, Keunwoo Choi, and Juhan Nam. 2025. TALKPLAY : Multimodal music recommendation with large language models. ArXiv preprint arXiv:2502.13713

    2025 arXiv

  9. [17]

    Hongliang Fei, Jingyuan Zhang, Xingxuan Zhou, Junhao Zhao, Xinyang Qi, and Ping Li. 2021. GemNN : Gating-enhanced multi-task neural networks with feature interaction learning for CTR prediction. In Proceedings of the 44th International ACM SIGIR Conference on Research and Deve...

    2021

  10. [18]

    Luciano Floridi and Massimo Chiriatti. 2020. GPT-3 : Its nature, scope, limits, and consequences. Minds and Machines, 30:681--694

    2020

  11. [19]

    Luke Gallagher, Ruey-Cheng Chen, Roi Blanco, and J Shane Culpepper. 2019. Joint optimization of cascade ranking models. In Proceedings of the twelfth ACM International Conference on Web Search and Data Mining, pages 15--23

    2019

  12. [20]

    Chongming Gao, Ruijun Chen, Shuai Yuan, Kexin Huang, Yuanqing Yu, and Xiangnan He. 2025. SPRec : Self-play to debias llm-based recommendation. In Proceedings of the ACM on Web Conference 2025, pages 5075--5084

    2025

  13. [21]

    Rohit Girdhar, Alaaeldin El-Nouby, Zhuang Liu, Mannat Singh, Kalyan Vasudev Alwala, Armand Joulin, and Ishan Misra. 2023. ImageBind : One embedding space to bind them all. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 15180--15190

    2023

  14. [22]

    Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, and 1 others. 2024. The Llama 3 herd of models. ArXiv preprint arXiv:2407.21783

    2024 arXiv

  15. [23]

    Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, and 1 others. 2025 a . DeepSeek-R1 : Incentivizing reasoning capability in llms via reinforcement learning. ArXiv preprint arXiv:2501.12948

    2025 arXiv

  16. [24]

    Hao Guo, Erpeng Xue, Lei Huang, Shichao Wang, Xiaolei Wang, Lei Wang, Jinpeng Wang, and Sheng Chen. 2025 b . Action is all you need: Dual-flow generative ranking network for recommendation. ArXiv preprint arXiv:2505.16752

    2025 arXiv

  17. [25]

    Xian Guo, Ben Chen, Siyuan Wang, Ying Yang, Chenyi Lei, Yuqing Ding, and Han Li. 2025 c . OneSug : The unified end-to-end generative framework for e-commerce query suggestion. ArXiv preprint arXiv:2506.06913

    2025 arXiv

  18. [26]

    Ruidong Han, Qianzhong Li, He Jiang, Rui Li, Yurou Zhao, Xiang Li, and Wei Lin. 2024. Enhancing CTR prediction through sequential recommendation pre-training: Introducing the SRP4CTR framework. In Proceedings of the 33rd ACM International Conference on Information and Knowledg...

    2024

  19. [27]

    Ruidong Han, Bin Yin, Shangyu Chen, He Jiang, Fei Jiang, Xiang Li, Chi Ma, Mincong Huang, Xiaoguang Li, Chunzhen Jing, and 1 others. 2025. MTGR : Industrial-scale generative recommendation framework in meituan. ArXiv preprint arXiv:2505.18654

    2025 arXiv

  20. [28]

    Xiangnan He, Lizi Liao, Hanwang Zhang, Liqiang Nie, Xia Hu, and Tat-Seng Chua. 2017. Neural collaborative filtering. In Proceedings of the 26th International Conference on World Wide Web, pages 173--182

    2017

  21. [29]

    Yupeng Hou, Jiacheng Li, Ashley Shin, Jinsung Jeon, Abhishek Santhanam, Wei Shao, Kaveh Hassani, Ning Yao, and Julian McAuley. 2025. Generating long semantic IDs in parallel for recommendation. ArXiv preprint arXiv:2506.05781

    2025 arXiv

  22. [30]

    Jun Hu, Wenwen Xia, Xiaolu Zhang, Chilin Fu, Weichang Wu, Zhaoxin Huan, Ang Li, Zuoli Tang, and Jun Zhou. 2024. Enhancing sequential recommendation via LLM -based semantic embedding learning. In Companion Proceedings of the ACM Web Conference 2024, pages 103--111

    2024

  23. [31]

    Lei Huang, Weitao Li, Chenrui Zhang, Jinpeng Wang, Xianchun Yi, and Sheng Chen. 2024. EXIT : An explicit interest transfer framework for cross-domain recommendation. In Proceedings of the 33rd ACM International Conference on Information and Knowledge Management, pages 4563--4570

    2024

  24. [32]

    Po-Sen Huang, Xiaodong He, Jianfeng Gao, Li Deng, Alex Acero, and Larry Heck. 2013. Learning deep structured semantic models for web search using clickthrough data. In Proceedings of the 22nd ACM International Conference on Information & Knowledge Management, pages 2333--2338

    2013

  25. [33]

    Shaohan Huang, Li Dong, Wenhui Wang, Yaru Hao, Saksham Singhal, Shuming Ma, Tengchao Lv, Lei Cui, Owais Khan Mohammed, Qiang Liu, and 1 others. 2023 a . Language is not all you need: Aligning perception with language models. ArXiv preprint arXiv:2302.14045

    2023 arXiv

  26. [34]

    Xu Huang, Defu Lian, Jin Chen, Liu Zheng, Xing Xie, and Enhong Chen. 2023 b . Cooperative retriever and ranker in deep recommenders. In Proceedings of the ACM Web Conference 2023, pages 1150--1161

    2023

  27. [35]

    Yanhua Huang, Yuqi Chen, Xiong Cao, Rui Yang, Mingliang Qi, Yinghao Zhu, Qingchang Han, Yaowei Liu, Zhaoyu Liu, Xuefeng Yao, and 1 others. 2025. Towards large-scale generative ranking. ArXiv preprint arXiv:2505.04180

    2025 arXiv

  28. [36]

    Jian Jia, Yipei Wang, Yan Li, Honggang Chen, Xuehan Bai, Zhaocheng Liu, Jian Liang, Quan Chen, Han Li, Peng Jiang, and 1 others. 2025. LEARN : Knowledge adaptation from large language model to recommendation for practical industrial application. In Proceedings of the AAAI Conf...

    2025

  29. [37]

    Junguang Jiang, Yanwen Huang, Bin Liu, Xiaoyu Kong, Ziru Xu, Han Zhu, Jian Xu, and Bo Zheng. 2025. Large language models are universal recommendation learners. ArXiv preprint arXiv:2502.03041

    2025 arXiv

  30. [38]

    Wang-Cheng Kang and Julian McAuley. 2018. Self-attentive sequential recommendation. In 2018 IEEE International Conference on Data Mining (ICDM), pages 197--206. IEEE

    2018

  31. [39]

    Saketh Reddy Karra and Theja Tulabandhula. 2024. InteraRec : Interactive recommendations using multimodal large language models. In Pacific-Asia Conference on Knowledge Discovery and Data Mining, pages 32--43. Springer

    2024

  32. [40]

    Kibum Kim, Sein Kim, Hongseok Kang, Jiwan Kim, Heewoong Noh, Yeonjun In, Kanghoon Yoon, Jinoh Oh, and Chanyoung Park. 2025. Image is all you need: Towards efficient and effective large language model-based recommender systems. ArXiv preprint arXiv:2503.06238

    2025

  33. [41]

    Wonjae Kim, Bokyung Son, and Ildoo Kim. 2021. ViLT : Vision-and-language transformer without convolution or region supervision. In International Conference on Machine Learning, pages 5583--5594. PMLR

    2021

  34. [42]

    Yehuda Koren, Robert Bell, and Chris Volinsky. 2009. Matrix factorization techniques for recommender systems. Computer, 42(8):30--37

    2009

  35. [43]

    Xuan Nhat Lam, Thuc Vu, Trong Duc Le, and Anh Duc Duong. 2008. Addressing cold-start problem in recommendation systems. In Proceedings of the 2nd International Conference on Ubiquitous Information Management and Communication, pages 208--211

    2008

  36. [44]

    Junnan Li, Ramprasaath Selvaraju, Akhilesh Gotmare, Shafiq Joty, Caiming Xiong, and Steven Chu Hong Hoi. 2021. Align before fuse: Vision and language representation learning with momentum distillation. Advances in Neural Information Processing Systems, 34:9694--9705

    2021

  37. [45]

    Xiaopeng Li, Xiangyang Li, Hao Zhang, Zhaocheng Du, Pengyue Jia, Yichao Wang, Xiangyu Zhao, Huifeng Guo, and Ruiming Tang. 2024 a . SyNeg : LLM -driven synthetic hard-negatives for dense retrieval. ArXiv preprint arXiv:2412.17250

    2024 arXiv

  38. [46]

    Yongqi Li, Xinyu Lin, Wenjie Wang, Fuli Feng, Liang Pang, Wenjie Li, Liqiang Nie, Xiangnan He, and Tat-Seng Chua. 2024 b . A survey of generative search and recommendation in the era of large language models. ArXiv preprint arXiv:2404.16924

    2024 arXiv

  39. [47]

    Jiayi Liao, Xiangnan He, Ruobing Xie, Jiancan Wu, Yancheng Yuan, Xingwu Sun, Zhanhui Kang, and Xiang Wang. 2024. RosePO : Aligning LLM -based recommenders with human values. ArXiv preprint arXiv:2410.12519

    2024 arXiv

  40. [48]

    Greg Linden, Brent Smith, and Jeremy York. 2003. Amazon. com recommendations: Item-to-item collaborative filtering. IEEE Internet Computing, 7(1):76--80

    2003

  41. [49]

    Chi Liu, Jiangxia Cao, Rui Huang, Kai Zheng, Qiang Luo, Kun Gai, and Guorui Zhou. 2024. KuaiFormer : Transformer-based retrieval at Kuaishou . ArXiv preprint arXiv:2411.10057

    2024 arXiv

  42. [50]

    Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2023 a . Improved baselines with visual instruction tuning. ArXiv preprint arXiv:2310.03744

    2023 arXiv

  43. [51]

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023 b . Visual instruction tuning. ArXiv preprint arXiv:2304.08485

    2023 arXiv

  44. [52]

    Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. RoBERTa : A robustly optimized BERT pretraining approach. ArXiv preprint arXiv:1907.11692

    2019 arXiv

  45. [53]

    Xinchen Luo, Jiangxia Cao, Tianyu Sun, Jinkai Yu, Rui Huang, Wei Yuan, Hezheng Lin, Yichen Zheng, Shiyao Wang, Qigen Hu, and 1 others. 2024. QARM : Quantitative alignment multi-modal recommendation at Kuaishou . ArXiv preprint arXiv:2411.11739

    2024 arXiv

  46. [54]

    Hanjia Lyu, Song Jiang, Hanqing Zeng, Yinglong Xia, Qifan Wang, Si Zhang, Ren Chen, Chris Leung, Jiajie Tang, and Jiebo Luo. 2024. https://doi.org/10.18653/v1/2024.findings-naacl.39 LLM-Rec : Personalized recommendation via prompting large language models . In Findings of the ...

    2024 doi

  47. [55]

    Julian McAuley. 2022. Personalized Machine Learning. Cambridge University Press

    2022

  48. [56]

    Aaron van den Oord, Yazhe Li, and Oriol Vinyals. 2018. Representation learning with contrastive predictive coding. ArXiv preprint arXiv:1807.03748

    2018 arXiv

  49. [57]

    Qi Pi, Guorui Zhou, Yujing Zhang, Zhe Wang, Lejian Ren, Ying Fan, Xiaoqiang Zhu, and Kun Gai. 2020. Search-based user interest modeling with lifelong sequential behavior data for click-through rate prediction. In Proceedings of the 29th ACM International Conference on Informat...

    2020

  50. [58]

    Junyan Qiu, Ze Wang, Fan Zhang, Zuowu Zheng, Jile Zhu, Jiangke Fan, Teng Zhang, Haitao Wang, and Xingxing Wang. 2025. One model to rank them all: Unifying online advertising with end-to-end learning. ArXiv preprint arXiv:2505.19755

    2025 arXiv

  51. [59]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, and 1 others. 2021. Learning transferable visual models from natural language supervision. In International Conference on Machine...

    2021

  52. [60]

    Rafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D Manning, Stefano Ermon, and Chelsea Finn. 2023. Direct preference optimization: Your language model is secretly a reward model. Advances in Neural Information Processing Systems, 36:53728--53741

    2023

  53. [61]

    Shashank Rajput, Nikhil Mehta, Anima Singh, Raghunandan Hulikal Keshavan, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Tran, Jonah Samost, and 1 others. 2023. Recommender systems with generative retrieval. Advances in Neural Information Processing Systems, 36:10299--10315

    2023

  54. [62]

    Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme. 2012. BPR : Bayesian personalized ranking from implicit feedback. ArXiv preprint arXiv:1205.2618

    2012 arXiv

  55. [63]

    Badrul Sarwar, George Karypis, Joseph Konstan, and John Riedl. 2001. Item-based collaborative filtering recommendation algorithms. In Proceedings of the 10th International Conference on World Wide Web, pages 285--295

    2001

  56. [64]

    Leheng Sheng, An Zhang, Yi Zhang, Yuxin Chen, Xiang Wang, and Tat-Seng Chua. 2024. Language representations can be what recommenders need: Findings and potentials. ArXiv preprint arXiv:2407.05441

    2024 arXiv

  57. [65]

    Jiaxi Tang and Ke Wang. 2018. Personalized top-n sequential recommendation via convolutional sequence embedding. In Proceedings of the Eleventh ACM International Conference on Web Search and Data Mining, pages 565--573

    2018

  58. [66]

    Gemini Team, Rohan Anil, Sebastian Borgeaud, Yonghui Wu, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, and 1 others. 2023. Gemini : a family of highly capable multimodal models. ArXiv preprint arXiv:2312.11805

    2023 arXiv

  59. [67]

    Chunqi Wang, Bingchao Wu, Zheng Chen, Lei Shen, Bing Wang, and Xiaoyi Zeng. 2025. Scaling transformers for discriminative recommendation via generative pretraining. ArXiv preprint arXiv:2506.03699

    2025 arXiv

  60. [68]

    Lidan Wang, Jimmy Lin, and Donald Metzler. 2011. A cascade ranking model for efficient ranked retrieval. In Proceedings of the 34th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 105--114

    2011

  61. [69]

    Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, and 1 others. 2024 a . Qwen2-vl : Enhancing vision-language model's perception of the world at any resolution. ArXiv preprint arXiv:2409.12191

    2024 arXiv

  62. [70]

    Yunli Wang, Zhiqiang Wang, Jian Yang, Shiyang Wen, Dongying Kong, Han Li, and Kun Gai. 2024 b . Adaptive neural ranking framework: Toward maximized business goal for cascade ranking systems. In Proceedings of the ACM Web Conference 2024, pages 3798--3809

    2024

  63. [71]

    Tianxin Wei, Ziwei Wu, Ruirui Li, Ziniu Hu, Fuli Feng, Xiangnan He, Yizhou Sun, and Wei Wang. 2020. Fast adaptation for cold-start collaborative filtering with meta-learning. In 2020 IEEE International Conference on Data Mining (ICDM), pages 661--670. IEEE

    2020

  64. [72]

    Hong-Jian Xue, Xinyu Dai, Jianbing Zhang, Shujian Huang, and Jiajun Chen. 2017. Deep matrix factorization models for recommender systems. In IJCAI, volume 17, pages 3203--3209. Melbourne, Australia

    2017

  65. [73]

    Jiawei Xue, Zhen Yang, Haitao Lin, Ziji Zhang, Luzhu Wang, Yikun Gu, Yao Xu, and Xin Li. 2025. HGCL : Hierarchical graph contrastive learning for user-item recommendation. ArXiv preprint arXiv:2505.19020

    2025 arXiv

  66. [74]

    Bencheng Yan, Shilei Liu, Zhiyuan Zeng, Zihao Wang, Yizhen Zhang, Yujin Yuan, Langming Liu, Jiaqi Liu, Di Wang, Wenbo Su, and 1 others. 2025. Unlocking scaling law in industrial recommendation systems with a three-step paradigm based large user model. ArXiv preprint arXiv:2502.08309

    2025

  67. [75]

    An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, and 1 others. 2025 a . Qwen3 technical report. ArXiv preprint arXiv:2505.09388

    2025 arXiv

  68. [76]

    Liu Yang, Fabian Paischer, Kaveh Hassani, Jiacheng Li, Shuai Shao, Zhang Gabriel Li, Yun He, Xue Feng, Nima Noorshams, Sem Park, and 1 others. 2024 a . Unifying generative and dense retrieval for sequential recommendation. ArXiv preprint arXiv:2411.18814

    2024 arXiv

  69. [77]

    Shenghao Yang, Weizhi Ma, Peijie Sun, Min Zhang, Qingyao Ai, Yiqun Liu, and Mingchen Cai. 2024 b . Common sense enhanced knowledge-based recommendation with large language model. In International Conference on Database Systems for Advanced Applications, pages 381--390. Springer

    2024

  70. [78]

    Yuhao Yang, Zhi Ji, Zhaopeng Li, Yi Li, Zhonglin Mo, Yue Ding, Kai Chen, Zijian Zhang, Jie Li, Shuanglong Li, and Lin Liu. 2025 b . https://arxiv.org/abs/2503.02453 Sparse meets dense: Unified generative recommendations with cascaded sparse-dense representations . Preprint, ar...

    2025 arXiv

  71. [79]

    Zhen Yang, Yingxue Zhang, Fandong Meng, and Jie Zhou. 2023. TEAL : Tokenize and embed all for multi-modal large language models. ArXiv preprint arXiv:2311.04589

    2023 arXiv

  72. [80]

    Jun Yin, Zhengxin Zeng, Mingzheng Li, Hao Yan, Chaozhuo Li, Weihao Han, Jianjin Zhang, Ruochen Liu, Allen Sun, Denvy Deng, and 1 others. 2024. Unleash LLMs potential for recommendation by coordinating twin-tower dynamic semantic token generator. ArXiv preprint arXiv:2409.09253

    2024 arXiv

  73. [81]

    Lijun Yu, Yong Cheng, Zhiruo Wang, Vivek Kumar, Wolfgang Macherey, Yanping Huang, David A Ross, Irfan Essa, Yonatan Bisk, Ming-Hsuan Yang, and 1 others. 2023. SPAE : Semantic pyramid autoencoder for multimodal generation with frozen llms. ArXiv preprint arXiv:2306.17842

    2023 arXiv

  74. [82]

    Zheng Yuan, Fajie Yuan, Yu Song, Youhua Li, Junchen Fu, Fei Yang, Yunzhu Pan, and Yongxin Ni. 2023. Where to go next for recommender systems? ID -vs. modality-based recommender models revisited. In Proceedings of the 46th International ACM SIGIR Conference on Research and Deve...

    2023

  75. [83]

    Neil Zeghidour, Alejandro Luebs, Ahmed Omran, Jan Skoglund, and Marco Tagliasacchi. 2021. SoundStream : An end-to-end neural audio codec. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 30:495--507

    2021

  76. [84]

    Jiaqi Zhai, Lucy Liao, Xing Liu, Yueming Wang, Rui Li, Xuan Cao, Leon Gao, Zhaojie Gong, Fangda Gu, Michael He, and 1 others. 2024. Actions speak louder than words: Trillion-parameter sequential transducers for generative recommendations. ArXiv preprint arXiv:2402.17152

    2024 arXiv

  77. [85]

    Chao Zhang, Haoxin Zhang, Shiwei Wu, Di Wu, Tong Xu, Xiangyu Zhao, Yan Gao, Yao Hu, and Enhong Chen. 2024 a . NoteLLM-2 : Multimodal large representation models for recommendation. ArXiv preprint arXiv:2405.16789

    2024 arXiv

  78. [86]

    Wenlin Zhang, Chuhan Wu, Xiangyang Li, Yuhao Wang, Kuicai Dong, Yichao Wang, Xinyi Dai, Xiangyu Zhao, Huifeng Guo, and Ruiming Tang. 2024 b . LLMTreeRec : Unleashing the power of large language models for cold-start recommendations. ArXiv preprint arXiv:2404.00702

    2024 arXiv

  79. [87]

    Yongfeng Zhang, Qingyao Ai, Xu Chen, and W Bruce Croft. 2017. Joint representation learning for top-n recommendation with heterogeneous information sources. In Proceedings of the 2017 ACM on Conference on Information and Knowledge Management, pages 1449--1458

    2017

  80. [88]

    Zihan Zhang, Meng Fang, Ling Chen, Mohammad-Reza Namazi-Rad, and Jun Wang. 2023. How do large language models capture the ever-changing world knowledge? a review of recent advances. ArXiv preprint arXiv:2310.07343

    2023 arXiv

  81. [89]

    Bowen Zheng, Yupeng Hou, Hongyu Lu, Yu Chen, Wayne Xin Zhao, Ming Chen, and Ji-Rong Wen. 2024. Adapting large language models by integrating collaborative semantics for recommendation. In 2024 IEEE 40th International Conference on Data Engineering (ICDE), pages 1435--1448. IEEE

    2024

  82. [90]

    Zuowu Zheng, Ze Wang, Fan Yang, Jiangke Fan, Teng Zhang, and Xingxing Wang. 2025. Beyond cascaded architectures: An end-to-end generative framework for industrial advertising. ArXiv e-prints, pages arXiv--2505

    2025

  83. [91]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list...

  84. [92]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.