Pith. sign in

REVIEW 3 major objections 4 minor 111 references

Unleashing the Power of Large Language Model for Denoising Recommendation

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Denoising recommendation is best done, the paper claims, by aligning LLM-generated preference and relation knowledge to recommendation targets through an information bottleneck, beating existing denoising methods on three datasets and two…

desk verdict LLaRD is a thoughtful LLM-knowledge-plus-information-bottleneck denoising framework with consistent gains, but the paper never shows its LLM-generated relation graph is built from the training split only, which could inflate the headline numbers. read the letter →

arxiv 2502.09058 v1 pith:3AFP5YJC submitted 2025-02-13 cs.IR

Shuyao Wang , Zhi Zheng , Yongduo Sui , Hui Xiong This is my paper · ORCID
classification cs.IR
keywords largelanguagemodelsdenoisingrecommendationinformationbottleneckchain-of-thoughtreasoningcollaborativefilteringimplicitfeedbackgraphneuralnetworksknowledge-enhanced
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

The paper aims to show that large language models can perform the hard part of recommendation denoising: deciding which observed clicks are noise and which unobserved items a user would genuinely like. It proposes LLaRD, which first uses an LLM to enrich user and item profiles into preference knowledge, then runs a chain-of-thought procedure over the user-item graph to produce relation knowledge—noise labels, similar-user links, and latent-interest items—and finally applies an information-bottleneck objective so that only LLM knowledge aligned with recommendation targets influences the denoised graph. The claimed payoff is practical: on Steam, Yelp, and Amazon-Book, with both GMF and LightGCN backbone models, LLaRD outperforms the comparison denoising methods, including the second-best method BOD by average relative gains of roughly 6.92% with GMF and 11.79% with LightGCN. A sympathetic reader would care because existing denoising methods lean on costly side information or on heuristics from observed interactions that fail when data is sparse; LLMs bring external world knowledge that can infer preferences beyond what the interaction log shows.

What carries the argument

The load-bearing object is the enriched interaction graph $G_{rel}$ built from the LLM's chain-of-thought outputs, together with the information-bottleneck training objective that decides how much of that generated knowledge survives. The chain-of-thought procedure is user-centric: it rates each interacted item High/Medium/Low, labels Low items that do not match the user's profile as noise, finds second-hop users with similar High-rated items, and proposes third-hop interest items, so the graph-level reasoning is decomposed into steps the LLM can handle. The objective maximizes mutual information between the denoised graph and both the recommendation targets and the LLM knowledge, while minimizing mutual information with the original noisy graph, using BPR loss, InfoNCE (a contrastive alignment loss), and HSIC (a kernel-based independence measure) as tractable surrogates. This lets the framework keep useful LLM knowledge and discard hallucinated or irrelevant knowledge during training.

What would settle it

Run LLaRD on a dataset with known injected noise and known held-out positives, then compute precision and recall of the LLM's $E_{Noise}$ against the injected noise and the overlap of $E_{Collab} \cup E_{Interests}$ with the held-out positives. Low precision on known noise, or overlap with test positives far above chance, would mean the gains come from label errors or leakage rather than denoising.

Watch

Extended reading notes

Core claim

LLaRD's central claim is that denoising should be treated as a knowledge-generation problem rather than only a loss-design problem. The LLM produces preference knowledge from profiles, keywords, and item text, and relation knowledge from a user-centric chain-of-thought over the interaction graph, yielding three edge sets: $E_{Noise}$ to remove, $E_{Collab}$ to add between similar users, and $E_{Interests}$ to add to latent-interest items. These form the enriched graph $G_{rel} = (U, I, (E - E_{Noise}) \cup E_{Collab} \cup E_{Interests})$. A learnable mask generator on the original graph produces the denoised graph $G'$, and the information-bottleneck objective $\max_{G'} I(R; G') + \alpha I(K_p, K_r; G') - \beta I(G'; G)$ is implemented with BPR loss, InfoNCE alignment losses, and an HSIC compression term. The paper reports that this consistently beats mainstream denoising baselines across all three datasets and both backbones, with average relative gains over BOD of about 6.92% on GMF and 11.79% on LightGCN.

Load-bearing premise

The method trusts the language model's judgments about which interactions are noise and which new links reflect real interest, and the paper does not check those judgments against ground truth.

Editorial extensions

If this is right

  • LLaRD outperforms the second-best baseline BOD by average relative gains of about 6.92% with GMF and 11.79% with LightGCN across Steam, Yelp, and Amazon-Book.
  • Removing the information-compression term (w/o MI_min) or the knowledge-alignment term (w/o MI_max) degrades performance, so both the bottleneck filtering and the LLM knowledge contribute to the result.
  • The LLM knowledge is generated in a one-time offline pass, keeping per-epoch training cost comparable to mainstream denoising methods despite the added graph and contrastive computations.
  • Cold-start users with sparse histories also improve, with LLaRD reporting consistent gains over baselines across all five sparsity groups on Amazon-Book and Steam.

Reading between the lines

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

  • A direct consequence the paper leaves implicit is that if the LLM's denoising labels are accurate, the cleaned graph can be consumed by any future backbone, decoupling denoising from architecture-specific loss design.
  • The paper does not check whether the LLM's added interest and collaborative edges overlap with held-out test positives; until such a leakage check is reported, the margin over BOD should be read with that caveat.
  • A natural extension is to feed the LLM's confidence scores into the learnable mask instead of using the labels only through the contrastive enriched view, letting the information bottleneck downweight uncertain judgments rather than treating them as binary facts.
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 / 4 minor

Summary. The paper proposes LLaRD, a framework that uses large language models (LLMs) to generate two types of denoising knowledge for collaborative filtering: preference knowledge derived from user/item text and relation knowledge derived from a user-centric chain-of-thought reasoning process over interaction graphs. The relation knowledge consists of identified noise edges, collaborative user links, and interest item links, which are combined into an enriched graph. A knowledge-enhanced denoising module then applies an information-bottleneck objective, combining BPR loss, InfoNCE alignment with preference and relation knowledge, and an HSIC-based compression term. Experiments on Steam, Yelp, and Amazon-Book with GMF and LightGCN backbones report consistent improvements over denoising baselines, with average relative gains over BOD of about 6.92% for GMF and 11.79% for LightGCN, plus ablation, robustness, and cold-start analyses. The code is publicly available.

Significance. If the central claim holds, LLaRD is a timely and practical contribution: it is one of the first systematic attempts to use LLM-generated knowledge specifically for data-level denoising in recommendation, and it combines semantic knowledge with graph-based collaborative reasoning in a way that goes beyond representation-level LLM alignment such as RLMRec. The release of code, the use of the all-rank evaluation protocol, and the inclusion of ablation, noise-robustness, and cold-start experiments are concrete strengths. However, the significance is conditional on two load-bearing points that the manuscript does not currently establish: the LLM-generated relation knowledge must be derived without access to held-out test/validation interactions, and the LLM's hard noise and interest labels must be reliable enough to guide denoising. Both points affect the interpretation of every reported gain in Table 1.

major comments (3)
  1. [Section 3.2.1, Eqs. (10)-(15), and Eq. (20)] The evaluation may be optimistic because the relation-knowledge generation is never explicitly restricted to the training split. In Step 3, N(2)_u is defined as the union over U_i_k for items rated High, where U_i_k denotes 'users who have interacted with item i_k'; in Step 4, N(3)_u is defined as the union over I_u_k for second-hop users, where I_u_k denotes 'items interacted by users'. These are taken over the interaction graph without specifying whether validation/test edges are excluded. Appendix C.1 only states that the data are divided into training, validation, and testing with a 3:1:1 ratio; it does not state that the LLM in Section 3.2 sees only the training part. If the original graph includes validation or test edges, then E_Interests can contain a test-positive item for user u, and that edge is inserted into G_rel in Eq. (15). After GNN propagation on G_rel, h_hat_u in Eq. (20) aggregates the test item's embedding, and the InfoNCE loss pulls h'_u toward h_hat_u, inflating the test-time score y'_ui for that held-out item. The same channel can carry validation positives through ECollab even without direct user-item edges. The paper must state clearly that all LLM knowledge generation (Steps 1-4) uses only the training subgraph, and ideally verify empirically that no generated interest/collab edge overlaps the validation/test positive set. This is load-bearing for the central claim of consistent gains.
  2. [Section 4.2, Table 1] The empirical support is weakened by the absence of any variance or significance information. Table 1 reports a single run per configuration, with no error bars, no standard deviations, and no significance tests. Several of the claimed gains are numerically small; for example, on Yelp with GMF, LLaRD's R@10 is 0.0708 versus BOD's 0.0706, and N@10 is 0.0578 versus 0.0574. A difference of 0.0002-0.0004 can easily be within run-to-run noise, especially for deep models with Gumbel-Softmax sampling. Because the paper's abstract and Section 4.2 state that LLaRD 'consistently outperforms' all baselines, the authors should report results over multiple random seeds with standard deviations and, where feasible, paired significance tests against the strongest baseline BOD.
  3. [Section 3.2.1, Step 2 and Eqs. (8)-(9)] The hard LLM noise labels are treated as ground truth without any validation. The set E_Noise is formed by the LLM's binary decisions and then removed from the enriched graph in Eq. (15); if the LLM mislabels genuine interactions as noise, the model is trained on a corrupted graph, and the InfoNCE alignment in Eq. (20) reinforces the error by aligning the denoised graph with a relation knowledge graph that already reflects that mislabeling. The paper provides no analysis of label quality: no comparison of E_Noise against known noisy interactions, no human evaluation of a sample of noise labels, and no robustness test with perturbed or partially incorrect labels. Since the denoising contribution is the central novelty, the reliability of these labels is a load-bearing assumption. I recommend adding a label-quality study (e.g., precision against a manually annotated sample, or an ablation that flips a fraction of noise labels) to support the claim that the LLM knowledge improves, rather than merely changes, the denoising signal.
minor comments (4)
  1. [Section 3.3.2, Eq. (18)] Equation (18) defines y'_ui but not y'_uj; the reader must infer that y'_uj = h'^T_u h'_j. Please add the definition for the negative item to make the BPR loss self-contained.
  2. [Section 3.1, Eq. (2)] The notation T_u = {T^1_u, T^2_u, ..., T^{|U|}_u} is confusing: the superscript k on T^k_u appears to index users rather than a sequence for a fixed user, and the size |U| suggests one entry per user. Please clarify the indexing, since the same symbol is later used for token-embedding matrices T_u.
  3. [Section 4.3, Table 2] The ablation study in Table 2 is reported only for Amazon-Book and Steam, not for Yelp. Since Table 1 includes Yelp and the method is claimed to work across all three datasets, adding the Yelp ablation (or explaining its omission) would strengthen the completeness of the study.
  4. [Appendix C.4] The hyperparameter description is incomplete: the values of alpha and beta from Eq. (25), the kernel bandwidths sigma_k and sigma_m from Eq. (21), and the contrastive temperature tau' from Eqs. (19)-(20) are not reported. These are free parameters that affect the reported results, so they should be listed for reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported gains are empirical and measured on held-out interactions, with no equation-level reduction to the LLM-generated knowledge.

full rationale

Walking the claimed derivation chain: preference knowledge (Eqs. 2-7) and relation knowledge (Eqs. 8-15) are produced by LLM inference over text and graph neighborhoods, then used in an information-bottleneck objective (Eq. 17) via standard alignment and compression losses (Eqs. 18-24). The headline results in Table 1 are held-out ranking metrics under the all-rank protocol (Appendix C.2), so they are not algebraically identical to any fitted parameter or to the LLM knowledge itself. No 'uniqueness theorem' or author-imported result is invoked to force the method; self-citations in the paper (e.g., [42-48,53,54]) are background and are not load-bearing for the central empirical claim. The one substantive concern is a possible evaluation-leakage risk: Section 3.2.2 defines the interaction graph over all r_ui=1 edges without explicitly stating that LLM knowledge generation is restricted to the 3:1:1 training split described in Appendix C.1, so third-hop 'interest' edges (Eqs. 12-13) could in principle include held-out positives that enter G_rel (Eq. 15) and steer h'_u through Eq. 20. That is a data-contamination/correctness risk, not a circularity: it is not an equation-level identity, and confirming it would require implementation details outside the paper. Therefore, no circular step is established and the circularity score is 0.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The load-bearing extras beyond the data and the base recommenders are two knowledge signals K_p and K_r produced by an external LLM, and the tuned weights in the loss. K_p and K_r are not physically postulated entities; they are outputs of a non-deterministic API. The hyperparameters alpha, beta, tau, tau_prime, and kernel bandwidths are chosen by hand or validation, and their values are only partially reported. The main unstated assumptions are the reliability of LLM hard labels and the validity of HSIC as a mutual-information surrogate.

free parameters (5)
  • alpha (alpha in Eq. 25) = not reported
    Balances the knowledge alignment losses L_prf and L_rel against BPR; value chosen by validation and not justified in the text.
  • beta (beta in Eq. 25) = not reported
    Controls the HSIC compression strength and is central to the information-bottleneck denoising objective.
  • Gumbel temperature tau = 0.0001
    Sharpness of the mask sampling in Eq. 16; affects whether the mask is nearly binary.
  • contrastive temperature tau_prime = tuned from {0.1, ..., 0.5}
    Scales InfoNCE logits in both preference and relation alignment losses, Eqs. 19 and 20.
  • kernel bandwidths sigma_k and sigma_m = not reported
    Used in Gaussian kernels for HSIC in Eq. 21; the choice changes the compression loss.
assumptions (5)
  • standard math The Information Bottleneck objective (Eq. 17) and its Lagrangian form are valid for the denoising problem, and HSIC approximates mutual information between G and G' (Eqs. 21-24).
    The paper relies on the IB framework and HSIC as a proxy for mutual information without proving the approximation tight.
  • domain assumption LLMs can infer user and item preferences beyond what observational interaction data shows, based on text profiles and world knowledge (Section 3.1).
    This is the core premise for preference knowledge; the inferred profiles are not directly validated.
  • domain assumption The CoT reasoning over 2-hop and 3-hop neighborhoods produces accurate noise labels, collaborative users, and interest items (Section 3.2.1).
    Hard labels from Eqs. 8-13 are injected into G_rel without ground-truth validation.
  • domain assumption BPR loss on all original positive interactions measures recommendation information in the denoised graph (Eq. 18).
    Standard BPR assumption; additionally, the text does not filter dropped edges out of the BPR training set.
  • domain assumption GNN message passing on the enriched graph G_rel, including user-user edges, is well-defined (Section 3.2.2 and Eq. 15).
    The paper does not specify how LightGCN's bipartite propagation handles the added user-user edges.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Unleashing the Power of Large Language Model for Denoising Recommendation." pith.science (2026). https://pith.science/paper/3AFP5YJC

@misc{pith2026250209058,
  author       = {Pith},
  title        = {Pith review of: Unleashing the Power of Large Language Model for Denoising Recommendation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3AFP5YJC}},
  note         = {Machine review of arXiv:2502.09058}
}
read the original abstract

Recommender systems are crucial for personalizing user experiences but often depend on implicit feedback data, which can be noisy and misleading. Existing denoising studies involve incorporating auxiliary information or learning strategies from interaction data. However, they struggle with the inherent limitations of external knowledge and interaction data, as well as the non-universality of certain predefined assumptions, hindering accurate noise identification. Recently, large language models (LLMs) have gained attention for their extensive world knowledge and reasoning abilities, yet their potential in enhancing denoising in recommendations remains underexplored. In this paper, we introduce LLaRD, a framework leveraging LLMs to improve denoising in recommender systems, thereby boosting overall recommendation performance. Specifically, LLaRD generates denoising-related knowledge by first enriching semantic insights from observational data via LLMs and inferring user-item preference knowledge. It then employs a novel Chain-of-Thought (CoT) technique over user-item interaction graphs to reveal relation knowledge for denoising. Finally, it applies the Information Bottleneck (IB) principle to align LLM-generated denoising knowledge with recommendation targets, filtering out noise and irrelevant LLM knowledge. Empirical results demonstrate LLaRD's effectiveness in enhancing denoising and recommendation accuracy.

Figures

Figures reproduced from arXiv: 2502.09058 by the authors.

Figure 1
Figure 1. (a) An intuitive example of learning user pref [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overview of the proposed LLaRD framework. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Impact comparison w.r.t. noise ratio in added in￾teraction data. The bars display Recall@20, while the curve shows the drop rate in performance. • w/o MI𝑚𝑖𝑛: Removes the process of minimizing mutual infor￾mation between the denoised and original interaction graph. • w/o MI𝑚𝑎𝑥 : Removes the process of maximizing mutual infor￾mation between the denoised graph and denoising knowledge. • w/o PK: Removes the integration … view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Recommendation performance over different cold [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: The CoT reasoning case of LLaRD. Adam optimizer without weight decay. We adopt the early stop technique based on the model’s performance on the validation set. To generate the preference knowledge and relation knowledge, we leverage the Qwen model (specifically, qwen-l…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

111 extracted references · 55 canonical work pages

  1. [1]

    Keqin Bao, Jizhi Zhang, Yang Zhang, Wenjie Wang, Fuli Feng, and Xiangnan He. 2023. Tallrec: An effective and efficient tuning framework to align large language model with recommendation. InProceedings of the 17th ACM Conference on Recommender Systems. 1007–1014

    2023

  2. [2]

    Zhi Bian, Shaojun Zhou, Hao Fu, Qihong Yang, Zhenqi Sun, Junjie Tang, Guiquan Liu, Kaikui Liu, and Xiaolong Li. 2021. Denoising user-aware memory network for recommendation. In Proceedings of the 15th ACM Conference on Recommender Systems. 400–410

    2021

  3. [3]

    Georg Buscher, Ludger Van Elst, and Andreas Dengel. 2009. Segment-level display time as implicit feedback: a comparison to eye tracking. In SIGIR. 67–74

    2009

  4. [4]

    Hanzhu Chen, Xu Shen, Qitan Lv, Jie Wang, Xiaoqi Ni, and Jieping Ye. 2024. SAC-KG: Exploiting Large Language Models as Skilled Automatic Constructors for Domain Knowledge Graphs. arXiv preprint arXiv:2410.02811 (2024)

    arXiv 2024

  5. [5]

    Jiaju Chen, Wang Wenjie, Chongming Gao, Peng Wu, Jianxiong Wei, and Qing- song Hua. 2024. Treatment Effect Estimation for User Interest Exploration on Recommender Systems. In SIGIR. 1861–1871

    2024

  6. [6]

    Liyi Chen, Zhi Li, Weidong He, Gong Cheng, Tong Xu, Nicholas Jing Yuan, and Enhong Chen. 2023. Entity summarization via exploiting description complemen- tarity and salience. IEEE Transactions on Neural Networks and Learning Systems 34, 11 (2023), 8297–8309

    2023

  7. [7]

    Liyi Chen, Ying Sun, Shengzhe Zhang, Yuyang Ye, Wei Wu, and Hui Xiong. 2024. Tackling Uncertain Correspondences for Multi-Modal Entity Alignment. InNIPS

    2024

  8. [8]

    Liyi Chen, Panrong Tong, Zhongming Jin, Ying Sun, Jieping Ye, and Hui Xiong

Show all 111 references
  1. [9]

    Sunhao Dai, Ninglu Shao, Haiyuan Zhao, Weijie Yu, Zihua Si, Chen Xu, Zhongx- iang Sun, Xiao Zhang, and Jun Xu. 2023. Uncovering chatgpt’s capabilities in recommender systems. In Proceedings of the 17th ACM Conference on Recom- mender Systems. 1126–1132

    2023

  2. [10]

    Jingtao Ding, Guanghui Yu, Xiangnan He, Fuli Feng, Yong Li, and Depeng Jin

  3. [11]

    Rod Ellis, Shawn Loewen, and Rosemary Erlam. 2006. Implicit and explicit corrective feedback and the acquisition of L2 grammar.Studies in second language acquisition 28, 2 (2006), 339–368

    2006

  4. [12]

    Wenqi Fan, Yao Ma, Qing Li, Yuan He, Eric Zhao, Jiliang Tang, and Dawei Yin

  5. [13]

    Xin Fu. 2010. Towards a model of implicit feedback for web search. Journal of the American Society for Information Science and Technology 61, 1 (2010), 30–49

    2010

  6. [14]

    Zeno Gantner, Lucas Drumond, Christoph Freudenthaler, and Lars Schmidt- Thieme. 2012. Personalized ranking for non-uniformly sampled items. In Pro- ceedings of KDD Cup 2011 . PMLR, 231–247

    2012

  7. [15]

    Graph neural networks for social recommendation. In WWW. 417–426

  8. [16]

    Yingqiang Ge, Mostafa Rahmani, Athirai Irissappane, Jose Sepulveda, James Caverlee, and Fei Wang. 2023. Automated data denoising for recommendation. arXiv preprint arXiv:2305.07070 (2023)

    2023 arXiv

  9. [17]

    Shijie Geng, Shuchang Liu, Zuohui Fu, Yingqiang Ge, and Yongfeng Zhang. 2022. Recommendation as language processing (rlp): A unified pretrain, personalized prompt & predict paradigm (p5). In Proceedings of the 16th ACM Conference on Recommender Systems. 299–315

    2022

  10. [18]

    Yunjun Gao, Yuntao Du, Yujia Hu, Lu Chen, Xinjun Zhu, Ziquan Fang, and Baihua Zheng. 2022. Self-guided learning to denoise for robust recommendation. In SIGIR. 1412–1422

    2022

  11. [19]

    Yongqiang Han, Hao Wang, Kefan Wang, Likang Wu, Zhi Li, Wei Guo, Yong Liu, Defu Lian, and Enhong Chen. 2024. Efficient Noise-Decoupling for Multi- Behavior Sequential Recommendation. In WWW. 3297–3306

    2024

  12. [20]

    Xiangnan He, Kuan Deng, Xiang Wang, Yan Li, Yongdong Zhang, and Meng Wang. 2020. Lightgcn: Simplifying and powering graph convolution network for recommendation. In SIGIR. 639–648

    2020

  13. [21]

    Michael Gutmann and Aapo Hyvärinen. 2010. Noise-contrastive estimation: A new estimation principle for unnormalized statistical models. In Proceedings of the thirteenth international conference on artificial intelligence and statistics . JMLR Workshop and Conference Proceeding...

    2010

  14. [22]

    Zhankui He, Zhouhang Xie, Rahul Jha, Harald Steck, Dawen Liang, Yesu Feng, Bodhisattwa Prasad Majumder, Nathan Kallus, and Julian McAuley. 2023. Large language models as zero-shot conversational recommenders. In CIKM. 720–730

    2023

  15. [23]

    Yifan Hu, Yehuda Koren, and Chris Volinsky. 2008. Collaborative filtering for implicit feedback datasets. In 2008 Eighth IEEE international conference on data mining. Ieee, 263–272

    2008

  16. [24]

    Zhuangzhuang He, Yifan Wang, Yonghui Yang, Peijie Sun, Le Wu, Haoyue Bai, Jinqi Gong, Richang Hong, and Min Zhang. 2024. Double Correction Framework for Denoising Recommendation. arXiv preprint arXiv:2405.11272 (2024)

    2024 arXiv

  17. [25]

    Yangqin Jiang, Yuhao Yang, Lianghao Xia, and Chao Huang. 2024. Diffkg: Knowl- edge graph diffusion model for recommendation. In WSDM. 313–321

    2024

  18. [26]

    Thorsten Joachims, Laura Granka, Bing Pan, Helene Hembrooke, and Geri Gay

  19. [27]

    Gawesh Jawaheer, Martin Szomszor, and Patty Kostkova. 2010. Comparison of implicit and explicit feedback from an online music recommendation service. In proceedings of the 1st international workshop on information heterogeneity and fusion in recommender systems . 47–51

    2010

  20. [28]

    Yohay Kaplan, Naama Krasne, Alex Shtoff, and Oren Somekh. 2021. Unbiased filtering of accidental clicks in verizon media native advertising. InProceedings of the 30th ACM International Conference on Information & Knowledge Management . 3878–3887

    2021

  21. [29]

    Farhan Khawar, Leonard Poon, and Nevin L Zhang. 2020. Learning the structure of auto-encoding recommenders. In WWW. 519–529

    2020

  22. [30]

    Hai-Dang Kieu, Minh Duc Nguyen, Thanh-Son Nguyen, and Dung D Le. 2024. Keyword-driven Retrieval-Augmented Large Language Models for Cold-start User Recommendations. arXiv preprint arXiv:2405.19612 (2024)

    2024 arXiv

  23. [31]

    Thorsten Joachims, Adith Swaminathan, and Tobias Schnabel. 2017. Unbiased learning-to-rank with biased feedback. In Proceedings of the tenth ACM interna- tional conference on web search and data mining . 781–789

    2017

  24. [32]

    Dawen Liang, Rahul G Krishnan, Matthew D Hoffman, and Tony Jebara. 2018. Variational autoencoders for collaborative filtering. In WWW. 689–698

    2018

  25. [33]

    Jiayi Liao, Sihang Li, Zhengyi Yang, Jiancan Wu, Yancheng Yuan, Xiang Wang, and Xiangnan He. 2024. Llara: Large language-recommendation assistant. In SIGIR. 1785–1795

    2024

  26. [34]

    Weilin Lin, Xiangyu Zhao, Yejing Wang, Yuanshao Zhu, and Wanyu Wang. 2023. Autodenoise: Automatic data instance denoising for recommendations. InWWW. 1003–1011

    2023

  27. [35]

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

    2009

  28. [36]

    Duc Tam Nguyen, Chaithanya Kumar Mummadi, Thi Phuong Nhung Ngo, Thi Hoai Phuong Nguyen, Laura Beggel, and Thomas Brox. 2019. Self: Learning to filter noisy labels with self-ensembling. arXiv preprint arXiv:1910.01842 (2019)

    2019 arXiv

  29. [37]

    Weike Pan and Li Chen. 2013. Gbpr: Group preference based bayesian personal- ized ranking for one-class collaborative filtering. In IJCAI

    2013

  30. [38]

    Xubin Ren, Wei Wei, Lianghao Xia, Lixin Su, Suqi Cheng, Junfeng Wang, Dawei Yin, and Chao Huang. 2024. Representation learning with large language models for recommendation. In WWW. 3464–3475

    2024

  31. [39]

    Junling Liu, Chao Liu, Peilin Zhou, Renjie Lv, Kang Zhou, and Yan Zhang

  32. [40]

    Amit Sharma, Jake M Hofman, and Duncan J Watts. 2015. Estimating the causal impact of recommendation systems from observational data. In Proceedings of the Sixteenth ACM Conference on Economics and Computation . 453–470

    2015

  33. [41]

    Florian Strub, Jeremie Mary, and Preux Philippe. 2015. Collaborative filtering with stacked denoising autoencoders and sparse inputs. In NIPS workshop on machine learning for eCommerce

    2015

  34. [42]

    Yongduo Sui, Wenyu Mao, Shuyao Wang, Xiang Wang, Jiancan Wu, Xiangnan He, and Tat-Seng Chua. 2024. Enhancing Out-of-distribution Generalization on Graphs via Causal Attention Learning. ACM TKDD 18, 5 (2024), 1–24

    2024

  35. [43]

    Yongduo Sui, Jie Sun, Shuyao Wang, Zemin Liu, Qing Cui, Longfei Li, and Xiang Wang. 2025. A Unified Invariant Learning Framework for Graph Classification. In SIGKDD

    2025

  36. [44]

    Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme

  37. [45]

    Yongduo Sui, Shuyao Wang, Jie Sun, Zhiyuan Liu, Qing Cui, Longfei Li, Jun Zhou, Xiang Wang, and Xiangnan He. 2024. A Simple Data Augmentation for Graph Classification: A Perspective of Equivariance and Invariance. ACM TKDD (2024)

    2024

  38. [46]

    Yongduo Sui, Xiang Wang, Jiancan Wu, Xiangnan He, and Tat-Seng Chua. 2022. Causal Attention for Interpretable and Generalizable Graph Classification. In SIGKDD. 1696–1705

    2022

  39. [47]

    Yongduo Sui, Qitian Wu, Jiancan Wu, Qing Cui, Longfei Li, Jun Zhou, Xiang Wang, and Xiangnan He. 2023. Unleashing the power of graph data augmentation on covariate distribution shift. NeurIPS 36 (2023)

    2023

  40. [48]

    Yong-Duo Sui, Xiang Wang, Tianlong Chen, Meng Wang, Xiang-Nan He, and Tat- Seng Chua. 2024. Inductive Lottery Ticket Learning for Graph Neural Networks. Journal of Computer Science and Technology 39, 6 (2024), 1223–1237

    2024

  41. [49]

    Qingyun Sun, Jianxin Li, Hao Peng, Jia Wu, Xingcheng Fu, Cheng Ji, and S Yu Philip. 2022. Graph structure learning with variational information bottleneck. In AAAI, Vol. 36. 4165–4174

    2022

  42. [50]

    Yongduo Sui, Caizhi Tang, Zhixuan Chu, Junfeng Fang, Yuan Gao, Qing Cui, Longfei Li, Jun Zhou, and Xiang Wang. 2024. Invariant Graph Learning for Causal Effect Estimation. In WWW. 2552–2562

    2024

  43. [51]

    Naftali Tishby and Noga Zaslavsky. 2015. Deep learning and the information bottleneck principle. In 2015 ieee information theory workshop (itw) . IEEE, 1–5

    2015

  44. [52]

    Gabriele Tolomei, Mounia Lalmas, Ayman Farahat, and Andrew Haines. 2019. You must have clicked on this ad by mistake! Data-driven identification of accidental clicks on mobile ads with applications to advertiser cost discounting and click- through rate prediction. Internationa...

    2019

  45. [53]

    Shuyao Wang, Yongduo Sui, Chao Wang, and Hui Xiong. 2024. Unleashing the Power of Knowledge Graph for Recommendation via Invariant Learning. In WWW. 3745–3755

    2024

  46. [54]

    Shuyao Wang, Yongduo Sui, Jiancan Wu, Zhi Zheng, and Hui Xiong. 2024. Dy- namic Sparse Learning: A Novel Paradigm for Efficient Recommendation. In WSDM. ACM

    2024

  47. [55]

    Wenjie Wang, Fuli Feng, Xiangnan He, Liqiang Nie, and Tat-Seng Chua. 2021. Denoising implicit feedback for recommendation. In WSDM. 373–381

    2021

  48. [56]

    Yatong Sun, Bin Wang, Zhu Sun, and Xiaochun Yang. 2021. Does Every Data In- stance Matter? Enhancing Sequential Recommendation by Eliminating Unreliable Data.. In IJCAI. 1579–1585

    2021

  49. [57]

    Yancheng Wang, Ziyan Jiang, Zheng Chen, Fan Yang, Yingxue Zhou, Eunah Cho, Xing Fan, Xiaojiang Huang, Yanbin Lu, and Yingzhen Yang. 2023. Rec- mind: Large language model powered agent for recommendation. arXiv preprint arXiv:2308.14296 (2023)

    2023 arXiv

  50. [58]

    Yu Wang, Xin Xin, Zaiqiao Meng, Joemon M Jose, Fuli Feng, and Xiangnan He

  51. [59]

    Zongwei Wang, Min Gao, Wentao Li, Junliang Yu, Linxin Guo, and Hongzhi Yin

  52. [60]

    Zitai Wang, Qianqian Xu, Zhiyong Yang, Xiaochun Cao, and Qingming Huang

  53. [61]

    Chunyu Wei, Jian Liang, Di Liu, and Fei Wang. 2022. Contrastive graph structure learning via information bottleneck for recommendation. NIPS 35 (2022), 20407– 20420

    2022

  54. [62]

    Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. 2022. Self-consistency improves chain of thought reasoning in language models. arXiv preprint arXiv:2203.11171 (2022)

    2022 arXiv

  55. [63]

    Wei Wei, Xubin Ren, Jiabin Tang, Qinyong Wang, Lixin Su, Suqi Cheng, Jun- feng Wang, Dawei Yin, and Chao Huang. 2024. Llmrec: Large language models with graph augmentation for recommendation. In Proceedings of the 17th ACM International Conference on Web Search and Data Mining...

    2024

  56. [64]

    Hongyi Wen, Longqi Yang, and Deborah Estrin. 2019. Leveraging post-click feedback for content recommendations. InProceedings of the 13th ACM Conference on Recommender Systems. 278–286

    2019

  57. [65]

    Jiancan Wu, Xiang Wang, Fuli Feng, Xiangnan He, Liang Chen, Jianxun Lian, and Xing Xie. 2021. Self-supervised graph learning for recommendation. In SIGIR. 726–735

    2021

  58. [66]

    Likang Wu, Zhi Zheng, Zhaopeng Qiu, Hao Wang, Hongchao Gu, Tingjia Shen, Chuan Qin, Chen Zhu, Hengshu Zhu, Qi Liu, et al . 2023. A Survey on Large Language Models for Recommendation. arXiv preprint arXiv:2305.19860 (2023)

    2023 arXiv

  59. [67]

    In SIGKDD

    Efficient bi-level optimization for recommendation denoising. In SIGKDD. 2502–2511

  60. [68]

    Yao Wu, Christopher DuBois, Alice X Zheng, and Martin Ester. 2016. Collab- orative denoising auto-encoders for top-n recommender systems. In WSDM. 153–162

    2016

  61. [69]

    Yunjia Xi, Weiwen Liu, Jianghao Lin, Xiaoling Cai, Hong Zhu, Jieming Zhu, Bo Chen, Ruiming Tang, Weinan Zhang, and Yong Yu. 2024. Towards open-world recommendation with knowledge augmentation from large language models. In Proceedings of the 18th ACM Conference on Recommender ...

    2024

  62. [70]

    Yuwei Xia, Ding Wang, Qiang Liu, Liang Wang, Shu Wu, and Xiao-Yu Zhang

  63. [71]

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. 2022. Chain-of-thought prompting elicits reasoning in large language models. NIPS 35 (2022), 24824–24837

    2022

  64. [72]

    Congxi Xiao, Jingbo Zhou, Jizhou Huang, Tong Xu, and Hui Xiong. 2023. Spatial heterophily aware graph neural networks. In SIGKDD. 2752–2763

    2023

  65. [73]

    Congxi Xiao, Jingbo Zhou, Jizhou Huang, Hengshu Zhu, Tong Xu, Dejing Dou, and Hui Xiong. 2023. A contextual master-slave framework on urban region graph for urban village detection. In ICDE. IEEE, 736–748

    2023

  66. [74]

    Congxi Xiao, Jingbo Zhou, Jizhou Huang, An Zhuo, Ji Liu, Haoyi Xiong, and Dejing Dou. 2021. C-watcher: A framework for early detection of high-risk neighborhoods ahead of covid-19 outbreak. In AAAI, Vol. 35. 4892–4900

    2021

  67. [75]

    Congxi Xiao, Jingbo Zhou, Yixiong Xiao, Jizhou Huang, and Hui Xiong. 2024. ReFound: Crafting a Foundation Model for Urban Region Understanding upon Language and Visual Foundations. In SIGKDD. 3527–3538

    2024

  68. [76]

    Tailin Wu, Hongyu Ren, Pan Li, and Jure Leskovec. 2020. Graph information bottleneck. NIPS 33 (2020), 20437–20448

    2020

  69. [77]

    Shenghao Yang, Weizhi Ma, Peijie Sun, Qingyao Ai, Yiqun Liu, Mingchen Cai, and Min Zhang. 2024. Sequential recommendation with latent relations based on large language model. In SIGIR. 335–344

    2024

  70. [78]

    Yonghui Yang, Le Wu, Richang Hong, Kun Zhang, and Meng Wang. 2021. En- hanced graph learning for collaborative filtering via mutual information maxi- mization. In SIGIR. 71–80

    2021

  71. [79]

    Yonghui Yang, Le Wu, Zihan Wang, Zhuangzhuang He, Richang Hong, and Meng Wang. 2024. Graph Bottlenecked Social Recommendation. In SIGKDD. 3853–3862

    2024

  72. [80]

    In Findings of the Association for Computational Linguistics ACL 2024

    Chain-of-History Reasoning for Temporal Knowledge Graph Forecasting. In Findings of the Association for Computational Linguistics ACL 2024 . 16144–16159

    2024

  73. [81]

    Yu Xia, Rui Wang, Xu Liu, Mingyan Li, Tong Yu, Xiang Chen, Julian McAuley, and Shuai Li. 2024. Beyond chain-of-thought: A survey of chain-of-x paradigms for llms. arXiv preprint arXiv:2404.15676 (2024)

    2024 arXiv

  74. [82]

    An Zhang, Yuxin Chen, Leheng Sheng, Xiang Wang, and Tat-Seng Chua. 2024. On generative agents in recommendation. In SIGIR. 1807–1817

    2024

  75. [83]

    Chi Zhang, Rui Chen, Xiangyu Zhao, Qilong Han, and Li Li. 2023. Denoising and prompt-tuning for multi-behavior recommendation. In WWW. 1355–1363

    2023

  76. [84]

    Chi Zhang, Qilong Han, Rui Chen, Xiangyu Zhao, Peng Tang, and Hongtao Song

  77. [85]

    Junjie Zhang, Ruobing Xie, Yupeng Hou, Wayne Xin Zhao, Leyu Lin, and Ji-Rong Wen. 2023. Recommendation as instruction following: A large language model empowered recommendation approach. arXiv preprint arXiv:2305.07001 (2023)

    2023 arXiv

  78. [86]

    Xin Xin, Xiangyuan Liu, Hanbing Wang, Pengjie Ren, Zhumin Chen, Jiahuan Lei, Xinlei Shi, Hengliang Luo, Joemon M Jose, Maarten de Rijke, et al . 2023. Improving implicit feedback-based recommendation through multi-behavior alignment. In SIGIR. 932–941

    2023

  79. [87]

    Wei Zhang, Chaoqun Wan, Yonggang Zhang, Yiu-ming Cheung, Xinmei Tian, Xu Shen, and Jieping Ye. 2024. Interpreting and Improving Large Language Models in Arithmetic Calculation. arXiv preprint arXiv:2409.01659 (2024)

    2024 arXiv

  80. [88]

    Yang Zhang, Keqin Bao, Ming Yan, Wenjie Wang, Fuli Feng, and Xiangnan He

  81. [89]

    Qian Zhao, Shuo Chang, F Maxwell Harper, and Joseph A Konstan. 2016. Gaze prediction for recommender systems. In Proceedings of the 10th ACM Conference on Recommender Systems. 131–138

    2016

  82. [90]

    Jia-Yu Yao, Kun-Peng Ning, Zhen-Hui Liu, Mu-Nan Ning, and Li Yuan. 2023. Llm lies: Hallucinations are not bugs, but features as adversarial examples. arXiv preprint arXiv:2310.01469 (2023)

    2023 arXiv

  83. [91]

    Junliang Yu, Hongzhi Yin, Xin Xia, Tong Chen, Lizhen Cui, and Quoc Viet Hung Nguyen. 2022. Are graph augmentations necessary? simple graph contrastive learning for recommendation. In SIGIR. 1294–1303

    2022

  84. [92]

    Zhi Zheng, Zhaopeng Qiu, Hui Xiong, Xian Wu, Tong Xu, Enhong Chen, and Xiangyu Zhao. 2022. DDR: Dialogue Based Doctor Recommendation for Online Medical Service. In SIGKDD. 4592–4600

    2022

  85. [93]

    Zhi Zheng, Zhaopeng Qiu, Tong Xu, Xian Wu, Xiangyu Zhao, Enhong Chen, and Hui Xiong. 2022. CBR: context bias aware recommendation for debiasing user modeling and click prediction. In WWW. 2268–2276

    2022

  86. [94]

    Zhi Zheng, Ying Sun, Xin Song, Hengshu Zhu, and Hui Xiong. 2023. Generative Learning Plan Recommendation for Employees: A Performance-aware Rein- forcement Learning Approach. In Proceedings of the 17th ACM Conference on Recommender Systems. 443–454

    2023

  87. [95]

    arXiv preprint arXiv:2403.04278 (2024)

    SSDRec: Self-Augmented Sequence Denoising for Sequential Recommen- dation. arXiv preprint arXiv:2403.04278 (2024)

    2024 arXiv

  88. [96]

    Zhi Zheng, Chao Wang, Tong Xu, Dazhong Shen, Penggang Qin, Xiangyu Zhao, Baoxing Huai, Xian Wu, and Enhong Chen. 2023. Interaction-aware drug package recommendation via policy gradient. ACM Transactions on Information Systems 41, 1 (2023), 1–32

    2023

  89. [97]

    Shengzhe Zhang, Liyi Chen, Chao Wang, Shuangli Li, and Hui Xiong. 2024. Temporal Graph Contrastive Learning for Sequential Recommendation. In AAAI, Vol. 38. 9359–9367

    2024

  90. [100]

    arXiv preprint arXiv:2406.03210 (2024)

    Text-like Encoding of Collaborative Information in Large Language Models for Recommendation. arXiv preprint arXiv:2406.03210 (2024)

    2024 arXiv

  91. [102]

    Yuyue Zhao, Jiancan Wu, Xiang Wang, Wei Tang, Dingxian Wang, and Maarten de Rijke. 2024. Let Me Do It For You: Towards LLM Empowered Recommendation via Tool Learning. In SIGIR. 1796–1806

    2024

  92. [103]

    Zhi Zheng, Zhaopeng Qiu, Xiao Hu, Likang Wu, Hengshu Zhu, and Hui Xiong

  93. [104]

    Generative job recommendations with large language model.arXiv preprint arXiv:2307.02157 (2023)

    2023 arXiv

  94. [108]

    Zhi Zheng, Chao Wang, Tong Xu, Dazhong Shen, Penggang Qin, Baoxing Huai, Tongzhu Liu, and Enhong Chen. 2021. Drug package recommendation via interaction-aware graph induction. In WWW. 1284–1295

    2021

  95. [110]

    Xinjun Zhu, Yuntao Du, Yuren Mao, Lu Chen, Yujia Hu, and Yunjun Gao. 2023. Knowledge-refined Denoising Network for Robust Recommendation. In SIGIR. 362–371. Unleashing the Power of Large Language Model for Denoising Recommendation WWW ’25, April 28–May 2, 2025, Sydney, NSW, Au...

    2023

  96. [111]

    light-hearted\

    During training, all methods are trained with a fixed batch size of 1024. We train all models using the learning rate 1e-3 with WWW ’25, April 28–May 2, 2025, Sydney, NSW, Australia. Shuyao Wang, Zhi Zheng, Yongduo Sui, & Hui Xiong User_ 1260 : Keywords: [small-town]/[humor]/ ...

    2025

  97. [2012]

    arXiv preprint arXiv:1205.2618 (2012)

    BPR: Bayesian personalized ranking from implicit feedback. arXiv preprint arXiv:1205.2618 (2012)

    2012 arXiv

  98. [2017]

    In SIGIR, Vol

    Accurately interpreting clickthrough data as implicit feedback. In SIGIR, Vol. 51. Acm New York, NY, USA, 4–11

  99. [2019]

    IEEE transactions on knowledge and data engineering 33, 2 (2019), 667–681

    Sampler design for bayesian personalized ranking by leveraging view data. IEEE transactions on knowledge and data engineering 33, 2 (2019), 667–681

    2019

  100. [2021]

    In Proceedings of the 29th ACM International Conference on Multimedia

    Implicit feedbacks are not always favorable: Iterative relabeled one-class collaborative filtering against noisy interactions. In Proceedings of the 29th ACM International Conference on Multimedia . 3070–3078

  101. [2022]

    Learning robust recommenders through cross-model agreement. In WWW. 2015–2025

    2015

  102. [2023]

    arXiv preprint arXiv:2304.10149 (2023)

    Is chatgpt a good recommender? a preliminary study. arXiv preprint arXiv:2304.10149 (2023)

    2023 arXiv

  103. [2024]

    Plan-on-Graph: Self-Correcting Adaptive Planning of Large Language Model on Knowledge Graphs. In NIPS

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.