Pith. sign in

REVIEW 3 major objections 3 minor 65 references

Divide-and-Conquer: Cold-Start Bundle Recommendation via Mixture of Diffusion Experts

T0 review · 3 major / 3 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A divide-and-conquer diffusion framework beats existing cold-start bundle recommenders.

desk verdict Novel framework, but the cold-start results hinge on an augmentation whose training data source is, read literally, the test split; the central claim does not hold as written. read the letter →

arxiv 2505.05035 v1 pith:GK72QNNM submitted 2025-05-08 cs.IR

Ming Li , Lin Li , Xiaohui Tao , Dong Zhang , Jimmy Xiangji Huang This is my paper · ORCID
classification cs.IR
keywords cold-startbundlerecommendationdivide-and-conquerdiffusionmodelmixtureofexpertscold-awaregatingrepresentationgenerationcollaborativefilteringaugmentation
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 paper proposes MoDiffE, a framework that treats cold-start bundle recommendation as a divide-and-conquer problem: split the cold-start difficulty by level (bundle and item) and by view (e.g., interaction views), solve each piece by training a diffusion model to generate the missing embedded representation directly, and combine the pieces with a cold-aware hierarchical mixture-of-experts gate. The authors claim this is the first work to formalize the dual-level multi-view complexity of the bundle cold-start problem, and that on three real-world datasets the framework outperforms existing cold-start recommenders, with up to 0.1027 absolute gain in Recall@20 in cold-start scenarios and up to 47.43% relative improvement in all-bundle scenarios. If right, it means new bundles without interaction history can be recommended immediately, without relying on hand-crafted content features for every view.

What carries the argument

The central mechanism is the divide-and-conquer pipeline with three load-bearing parts. First, a 'prior-embedding model' (a graph collaborative filtering backbone) produces embedded representations, which fail for feature-missing cold entities. Second, a denoising diffusion probabilistic model per view is trained to map warm embedded representations back to themselves with an L2 loss, and at inference denoises from a similarity-based anchor representation (top-$n$ most compositionally similar warm bundles' mean representation) to output a 'diffusion representation' for any input bundle, deterministic rather than diverse. Third, a cold-aware hierarchical MoE combines experts: view-layer gating (Softmax over cold-aware features such as interaction counts) fuses embedded and diffusion representations per view, and output-layer gating (Tanh) fuses view predictions. The cold-start gating augmentation creates pseudo cold bundles by interpolating pairs of cold-bundle representations (with cold-aware feature set to 0), enabling gating training for entities absent from the training set.

What would settle it

Inspect the released code or re-run the cold-start protocol with the interpolation sampling restricted to bundles that are never used for evaluation; if Recall@20 in the cold-start scenario falls toward the zero values reported for the no-augmentation ablation, the central empirical claim fails. A second check: count how many of the interpolated pairs in Stage 3 come from the validation or test bundle IDs.

Watch

Extended reading notes

Core claim

The paper's central claim is that the bundle cold-start problem decomposes into independent sub-problems—one per level and view—and that all of them share one root failure: a prior-embedding model, given missing features, produces no usable representation ($r = f^v_\theta(\emptyset) = \emptyset$). MoDiffE overcomes this by training a denoising diffusion model on the representations of warm bundles to capture their distribution, then, at inference, denoising from a similarity-based anchor representation of the cold bundle to generate a diffusion representation without requiring the missing features. A cold-aware hierarchical mixture of experts fuses embedded and diffusion representations per view (Softmax gating on cold-aware features like interaction counts) and then fuses view predictions (Tanh gating), so different cold-start situations get different expert mixes. A cold-start gating augmentation synthesizes pseudo cold bundles by linear interpolation so the gates can learn to route cold entities during training. The paper reports that this system outperforms the existing state of the art on Youshu, NetEase, and iFashion across cold-start, all-bundle, and warm-start scenarios.

Load-bearing premise

The cold-start gating augmentation samples bundle pairs from the set of bundles with no user interactions, but the paper never states where those cold bundles come from at training time; if they come from the validation or test split, the reported cold-start gains would be inflated by training the gates on the evaluation distribution.

Editorial extensions

If this is right

  • Cold bundles with no interaction history can still receive personalized ranking, because the diffusion expert generates bundle-level and item-level representations without relying on missing features.
  • The same framework can be dropped into other dual-level multi-view bundle models: any prior-embedding backbone can serve as the embedding expert, with diffusion experts trained on its warm representations.
  • Because the gating is conditioned on cold-aware features, different bundles in different cold-start situations (cold at bundle level, cold at item level, or both) receive different weightings of embedded versus generated representations.
  • In the all-bundle scenario, the model improves overall Recall@20 and NDCG@20 while shifting some hits from warm bundles to cold bundles, a fairness trade-off the paper's hit analysis documents.
  • The cold-start gating augmentation is necessary for the framework's cold-start performance: without it, the ablation shows Recall@20 and NDCG@20 drop to 0 in the cold-start scenario on all three datasets.

Reading between the lines

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

  • One consequence the paper leaves implicit: the augmentation step requires access to genuine cold bundles during training, so a strict deployment to truly unseen bundles would need either a source of cold bundles or an augmentation scheme that synthesizes them from warm bundles; the paper does not demonstrate that warm-only augmentation preserves the gains.
  • A testable extension would be to replace the diffusion expert with a non-diffusion denoiser (for example, an MLP trained with the same L2 objective and anchor sampling) to isolate whether multi-step denoising, rather than representation generation in general, drives the reported improvement.
  • The divide-conquer-combine pattern is not tied to interaction views; the same diffusion-plus-gating structure could be applied to content or knowledge-graph views, where feature-missing entities also produce empty prior-embedding representations, though the paper only sketches this as future work.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 3 minor

Summary. The paper proposes MoDiffE, a framework for cold-start bundle recommendation that divides the problem into sub-problems by level and view, generates bundle/item representations with diffusion models, and combines them via a cold-aware hierarchical mixture-of-experts (MoE) gating mechanism. A multi-stage decoupled training pipeline is used, and a cold-start gating augmentation is introduced to synthesize pseudo cold bundles by interpolation. Experiments on Youshu, NetEase, and iFashion claim state-of-the-art results in cold-start, all-bundle, and warm-start scenarios, with up to 0.1027 absolute Recall@20 gain in cold-start and 47.43% relative improvement in all-bundle settings.

Significance. If the empirical results were sound, the paper would make a meaningful contribution by proposing a new architecture for cold-start bundle recommendation that explicitly handles dual-level multi-view complexity, and by demonstrating the usefulness of diffusion models and mixture-of-experts for this task. The divide-and-conquer framing is conceptually appealing, and the multi-stage decoupled training is a practical engineering contribution. However, the central empirical claim rests on a training-data provenance issue that, as written, appears to involve evaluation-set leakage; the absence of error bars, significance tests, and a reported backbone baseline further weakens the evidence. Because the main contribution is empirical state-of-the-art performance, the validity of that claim is the decisive factor, and it is not established by the manuscript.

major comments (3)
  1. [Section 4.4.2, Eq. 23, Algorithm 3] The source of the cold bundle set B_c^{B-int} used for pseudo-cold bundle synthesis is unspecified and, by the paper's own definitions, cannot be part of the training split. Section 3.1 defines B_c^{B-int} as bundles with no historical user interactions, and Section 5.1.3 states that such bundles 'only appear after a specific data splitting' — meaning they are not in the training set. If the sampled bundle pairs come from the validation or test portions of the 7:1:2 split, then the gating networks are trained on the evaluation distribution, which would explain the large cold-start gains and invalidate the comparison against CoHeat and all other baselines. The ablation in Table 4 shows that without this augmentation, cold-start Recall@20 and NDCG@20 are exactly 0 on all datasets, so the augmentation is the entire source of cold-start competence. The manuscript must specify the exact provenance of the sampled cold bundles and, if they are drawn from validation/test, the experiments must be redesigned.
  2. [Section 5.2, Table 3] No error bars, multiple seeds, or significance tests are reported, so the claim that MoDiffE 'significantly outperforms state-of-the-art baselines' is not statistically supported. This matters particularly where the absolute differences are small, e.g., on NetEase cold-start Recall@20 MoDiffE is 0.0235 versus CoHeat's 0.0191, and on iFashion warm-start Recall@20 the difference is 0.1182 versus 0.1156. At minimum, standard deviations over several random seeds and paired significance tests (e.g., paired t-test or Wilcoxon) are required for the central claims.
  3. [Section 5.1.5 and Table 3] Hyperparameters are tuned per scenario and per dataset (e.g., interpolation ratio η = 0, 0.3, 0.5 for warm-start, all-bundle, and cold-start scenarios; noise schedule; learning rates; top-n for similarity-based sampling). Without a clear validation-based selection protocol that is held fixed before evaluating on the test set, the reported results risk overfitting to the test set. Additionally, the backbone CrossCBR is not reported in Table 3, so the reader cannot verify the claimed improvements over the prior-embedding model from which MoDiffE is built.
minor comments (3)
  1. [Section 5.4.1, Figure 3/4] The captions and panel labels for Figures 3 and 4 appear scrambled: the same sub-figure images (a-d) are duplicated across the two figures, and the caption of Figure 4 lists panels that do not match the text description. This makes the hyperparameter experiments hard to interpret.
  2. [Throughout] There are repeated typos, e.g., 'MoDiifE' in Section 5.2, 'CoHEAT' in the iFashion table, 'dependency on' in Section 5.3, and an incomplete sentence 'However, However, this definition contributes by explicitly emphasizes...' in Section 3.1. The paper would benefit from a careful proofread.
  3. [Section 4.4.1, Eq. 22] Equation 22 defines S with notation '<b_x, b_y>' and '<b'_x, b'_y>' that is not explained; the reader cannot tell whether negative samples are also interpolated or only positives. Please clarify the notation and the construction of negative samples for the BPR loss on the augmented set.

Circularity Check

1 steps flagged · score 6.0 of 10

Cold-start gating augmentation trains on the evaluation split, making the reported cold-start gains fitted rather than predicted.

  1. fitted input called prediction [Section 4.4.2 (Eq. 22-23, Algorithm 3 Stage 3) with Section 5.1.3]
    "Note that the original dataset does not contain bundle-level cold bundles (i.e. bundles without user interaction). In the experiment, it only appears after a specific data splitting. ... given a bundle pair <b_x,b_y> sampled from the cold bundle set B^{B-int}_c, we use a linear interpolation method [49, 58] to construct the representations and cold-aware feature of the pseudo cold bundle b_xy."

    By the paper's own definitions, B_c^{B-int} is the set of bundles with no historical user interactions (Section 3.1), and the training interactions X are restricted to warm bundles. Section 5.1.3 states that the original datasets contain no such bundles and that they 'only appear after a specific data splitting.' Hence the cold-bundle pairs sampled in Algorithm 3/Stage 3 and Eq. 22-23 must come from the validation/test portions of the 7:1:2 split. The gating networks are therefore trained on interpolations of evaluation bundles, and Table 3 then reports cold-start Recall@20/NDCG@20 for the same cold-bundle population. The cold-start 'prediction' is not an independent forecast; it is a fitted output on the evaluation distribution.

full rationale

The diffusion-model and MoE machinery is not equation-level circular: diffusion experts are trained on warm embedded representations and condition on bundle-item features, and the hierarchical gating combines independently trained expert outputs. The load-bearing circularity is confined to the cold-start gating augmentation. Because bundle-level cold bundles are, by the paper's own definitions, absent from the original training data and appear only after the 7:1:2 split, the set B_c^{B-int} sampled in Algorithm 3 can only be the validation/test cold bundles. Training the gating networks on interpolations of those bundles and then reporting cold-start metrics on the same population means the headline cold-start advantage is an artifact of fitting to the evaluation distribution, not a prediction for unseen bundles. The ablation confirms that all cold-start competence vanishes without this augmentation. Warm-start and some all-bundle results retain independent content, and the authors' other self-citations are not load-bearing, so the overall circularity score is 6 rather than higher.

Assumptions & free parameters 7 free parameters · 6 assumptions · 1 invented entities

The framework rests on several domain assumptions about decomposability and distribution transfer, plus a set of tuned hyperparameters. No formal proofs are provided; the validity of the empirical claim is the focus.

free parameters (7)
  • top_n = 5
    Number of similar warm bundles averaged to form the inference anchor (Section 5.1.5). Figure 3(c) shows per-dataset optima of 10 (Youshu) and 1 (NetEase), so 5 is a hand-chosen compromise.
  • Beta shape alpha = 0.9
    Shape parameter for the Beta distribution used in interpolation (Section 5.1.5). Figure 4(c,d) shows low sensitivity in most cases but it is tuned on the Youshu cold-start scenario.
  • interpolation sampling ratio eta = 0, 0.3, 0.5
    Ratio of augmented pseudo cold bundles to original training data, set per scenario (warm, all-bundle, cold-start) in Section 5.1.5. This controls how much synthetic cold data the gating sees.
  • diffusion forward steps T = 500
    Number of noising steps, tuned from {100,200,500,1000,2000,5000} in Section 5.4.1.
  • diffusion inference steps T' = 20
    Number of denoising steps, tuned from {1,5,10,20,30,40} in Section 5.4.1, with DPM-Solver.
  • noise schedule = linear (chosen)
    Tuned among linear, cosine linear, and exp schedules (Section 5.1.5). The final choice per dataset is not reported.
  • learning rate and L2 weight = searched
    Learning rates and L2 regularization for diffusion and gating networks are searched over the ranges listed in Section 5.1.5.
assumptions (6)
  • domain assumption The bundle cold-start problem decomposes into independent sub-problems by level and view whose solutions can be combined additively.
    Section 4.1.1 states the division is into independent but similar sub-problems; no proof or empirical test that interactions between sub-problems are negligible.
  • domain assumption Diffusion models trained to denoise warm bundle/item embedded representations can generate valid representations for cold bundles with missing interaction features.
    Section 4.2 assumes the warm representation distribution is informative for cold bundles; no out-of-distribution analysis is provided.
  • domain assumption The mean of top-n similar warm bundle representations is a good anchor for generating a cold bundle's representation.
    Section 4.2.5, Eq. 14, relies on item-composition similarity being predictive of representation similarity for cold bundles.
  • domain assumption Item features (the condition c) are available and informative for cold bundles and cold items.
    Section 4.2.3 treats items within a bundle as features pretrained from bundle-item affiliations; cold items may have uninformative initial embeddings.
  • ad hoc to paper Interpolated pseudo cold bundles approximate the distribution of real cold bundles closely enough to train gating networks.
    Section 4.4.2, Eq. 23, uses mixup-style interpolation of representations and assumes it simulates cold bundles without negative impact on the learned distribution.
  • standard math Standard DDPM mathematics (forward/reverse Gaussian processes, reparameterization, simplified loss) holds as in Ho et al. 2020.
    Sections 4.2.2-4.2.4 use the DDPM formulation without new derivation.
invented entities (1)
  • Pseudo cold bundle
    purpose: Synthetic training instances for gating networks, created by interpolating representations of two cold bundles and setting the cold-aware feature to 0.
    Not a physical entity; it is a data augmentation construct. Its source distribution is ambiguous, and there is no independent evidence outside the paper that it behaves like real cold bundles.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Divide-and-Conquer: Cold-Start Bundle Recommendation via Mixture of Diffusion Experts." pith.science (2026). https://pith.science/paper/GK72QNNM

@misc{pith2026250505035,
  author       = {Pith},
  title        = {Pith review of: Divide-and-Conquer: Cold-Start Bundle Recommendation via Mixture of Diffusion Experts},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GK72QNNM}},
  note         = {Machine review of arXiv:2505.05035}
}
read the original abstract

Cold-start bundle recommendation focuses on modeling new bundles with insufficient information to provide recommendations. Advanced bundle recommendation models usually learn bundle representations from multiple views (e.g., interaction view) at both the bundle and item levels. Consequently, the cold-start problem for bundles is more challenging than that for traditional items due to the dual-level multi-view complexity. In this paper, we propose a novel Mixture of Diffusion Experts (MoDiffE) framework, which employs a divide-and-conquer strategy for cold-start bundle recommendation and follows three steps:(1) Divide: The bundle cold-start problem is divided into independent but similar sub-problems sequentially by level and view, which can be summarized as the poor representation of feature-missing bundles in prior-embedding models. (2) Conquer: Beyond prior-embedding models that fundamentally provide the embedded representations, we introduce a diffusion-based method to solve all sub-problems in a unified way, which directly generates diffusion representations using diffusion models without depending on specific features. (3) Combine: A cold-aware hierarchical Mixture of Experts (MoE) is employed to combine results of the sub-problems for final recommendations, where the two models for each view serve as experts and are adaptively fused for different bundles in a multi-layer manner. Additionally, MoDiffE adopts a multi-stage decoupled training pipeline and introduces a cold-start gating augmentation method to enable the training of gating for cold bundles. Through extensive experiments on three real-world datasets, we demonstrate that MoDiffE significantly outperforms existing solutions in handling cold-start bundle recommendation. It achieves up to a 0.1027 absolute gain in Recall@20 in cold-start scenarios and up to a 47.43\% relative improvement in all-bundle scenarios.

Figures

Figures reproduced from arXiv: 2505.05035 by the authors.

Figure 1
Figure 1. (a) Bundle-level interaction-view representation learning. (b) Item-level interaction-view representation [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The MoDiffE framework the bundle level, but the root cause of its cold start is still in representation learning. Based on the above observation, the cold-start sub-problem across different levels and views can be summarized as a unified sub-problem that the poor representation of feature-missing bundles/items in the prior-embedding model. This can be formulated as: 𝑟 = 𝑓 𝑣 𝜃 (ℎ = ∅) = ∅ (7) This problem is shown in… view at source ↗
Figure 3
Figure 3. Impact of diffusion model hyperparameters on Recall@20 in cold-start scenario on Youshu and NetEase: [PITH_FULL_IMAGE:figures/full_fig_p020_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Impact of the cold-start gating augmentation method’s hyperparameters on Recall@20 in cold-start [PITH_FULL_IMAGE:figures/full_fig_p021_4.png]
Figure 5
Figure 5. Figure 5: Distributions of bundles/items in interaction views under different experts in the cold-start scenario of [PITH_FULL_IMAGE:figures/full_fig_p022_5.png]
Figure 6
Figure 6. Figure 6: View-layer gating and output-layer gating of warm bundle #494, cold bundle #1471, cold item #9090, [PITH_FULL_IMAGE:figures/full_fig_p023_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

65 extracted references · 61 canonical work pages

  1. [1]

    Oren Barkan, Noam Koenigstein, Eylon Yogev, and Ori Katz. 2019. CB2CF: A Neural Multiview Content-to-Collaborative Filtering Model for Completely Cold Item Recommendations. InProceedings of the 13th ACM Conference on Recommender Systems (RecSys ’19). ACM, 228–236

    work page 2019

  2. [2]

    Da Cao, Liqiang Nie, Xiangnan He, Xiaochi Wei, Shunzhi Zhu, and Tat-Seng Chua. 2017. Embedding Factorization Models for Jointly Recommending Items and User Generated Lists. InProceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR ’17’). ACM, 585–594

    work page 2017

  3. [3]

    Dong-Kyu Chae, Jin-Soo Kang, Sang-Wook Kim, and Jaeho Choi. 2019. Rating Augmentation with Generative Adversarial Networks towards Accurate Collaborative Filtering. InThe World Wide Web Conference (WWW ’19). ACM, 2616–2622

    work page 2019

  4. [4]

    Jianxin Chang, Chen Gao, Xiangnan He, Depeng Jin, and Yong Li. 2020. Bundle Recommendation with Graph Convolutional Networks. InProceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR ’20). ACM, 1673–1676

    work page 2020

  5. [5]

    Hao Chen, Zefan Wang, Feiran Huang, Xiao Huang, Yue Xu, Yishi Lin, Peng He, and Zhoujun Li. 2022. Generative Adversarial Framework for Cold-Start Item Recommendation. InProceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR ’22). ACM, 2565–2571

    work page 2022

  6. [6]

    Liang Chen, Yang Liu, Xiangnan He, Lianli Gao, and Zibin Zheng. 2019. Matching User with Item Set: Collaborative Bundle Recommendation with Deep Attention Network. InProceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence (IJCAI ’19). International Joint Conferences on Artificial Intelligence Organization, 2095–2101

    work page 2019

  7. [7]

    Weiss, Mohammad Norouzi, and William Chan

    Nanxin Chen, Yu Zhang, Heiga Zen, Ron J. Weiss, Mohammad Norouzi, and William Chan. 2020. WaveGrad: Es- timating Gradients for Waveform Generation. InInternational Conference on Learning Representations (ICLR ’20). OpenReview.net

    work page 2020

  8. [8]

    Wen Chen, Pipei Huang, Jiaming Xu, Xin Guo, Cheng Guo, Fei Sun, Chao Li, Andreas Pfadler, Huan Zhao, and Binqiang Zhao. 2019. POG: Personalized Outfit Generation for Fashion Recommendation at Alibaba iFashion. InProceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD ’19). ACM, 2662–2670

    work page 2019

Show all 65 references
  1. [9]

    Prafulla Dhariwal and Alex Nichol. 2021. Diffusion Models Beat GANs on Image Synthesis. InProceedings of the 35th International Conference on Neural Information Processing Systems (NeurIPS ’21). Curran Associates Inc., 8780–8794

    2021

  2. [10]

    Xiaoyu Du, Xiang Wang, Xiangnan He, Zechao Li, Jinhui Tang, and Tat-Seng Chua. 2020. How to Learn Item Representation for Cold-Start Multimedia Recommendation?. InProceedings of the 28th ACM International Conference on Multimedia (MM ’20). ACM, 3469–3477

    2020

  3. [11]

    Xidong Feng, Chen Chen, Dong Li, Mengchen Zhao, Jianye Hao, and Jun Wang. 2021. CMML: Contextual Modulation Meta Learning for Cold-Start Recommendation. InProceedings of the 30th ACM International Conference on Information & Knowledge Management (CIKM ’21). ACM, 484–493

    2021

  4. [12]

    Chen Gao, Yu Zheng, Nian Li, Yinfeng Li, Yingrong Qin, Jinghua Piao, Yuhan Quan, Jianxin Chang, Depeng Jin, Xiangnan He, and Yong Li. 2023. A Survey of Graph Neural Networks for Recommender Systems: Challenges, Methods, and Directions. 1, 1 (2023), 3:1–3:51

    2023

  5. [13]

    Shansan Gong, Mukai Li, Jiangtao Feng, Zhiyong Wu, and Lingpeng Kong. 2023. DiffuSeq: Sequence to Sequence Text Generation with Diffusion Models. InThe Eleventh International Conference on Learning Representations (ICLR ’23). OpenReview.net

    2023

  6. [14]

    Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron C

    Ian J. Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron C. Courville, and Yoshua Bengio. 2014. Generative Adversarial Nets. InProceedings of the 28th International Conference on Neural Information Processing Systems (NeurIPS ’14). ...

    2014

  7. [15]

    Jyotirmoy Gope and Sanjay Kumar Jain. 2017. A Survey on Solving Cold Start Problem in Recommender Systems. In 2017 International Conference on Computing, Communication and Automation (ICCCA ’17). 133–138

    2017

  8. [16]

    Xiangnan He, Kuan Deng, Xiang Wang, Yan Li, YongDong Zhang, and Meng Wang. 2020. LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation. InProceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval (...

    2020

  9. [17]

    Xiangnan He, Lizi Liao, Hanwang Zhang, Liqiang Nie, Xia Hu, and Tat-Seng Chua. 2017. Neural Collaborative Filtering. InProceedings of the 26th International Conference on World Wide Web (WWW ’17). International World Wide Web Conferences Steering Committee, 173–182. J. ACM, Vo...

    2017

  10. [18]

    Yun He, Jianling Wang, Wei Niu, and James Caverlee. 2019. A Hierarchical Self-Attentive Model for Recommending User-Generated Item Lists. InProceedings of the 28th ACM International Conference on Information and Knowledge Management (CIKM ’19). ACM, 1481–1490

    2019

  11. [19]

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. 2020. Denoising Diffusion Probabilistic Models. InProceedings of the 34th International Conference on Neural Information Processing Systems (NeurIPS ’20). Curran Associates Inc., 6840–6851

    2020

  12. [20]

    Hyunsik Jeon, Jong-eun Lee, Jeongin Yun, and U Kang. 2024. Cold-Start Bundle Recommendation via Popularity-Based Coalescence and Curriculum Heating. InProceedings of the ACM Web Conference 2024 (WWW ’24). ACM, 3277–3286

    2024

  13. [21]

    Yangqin Jiang, Lianghao Xia, Wei Wei, Da Luo, Kangyi Lin, and Chao Huang. 2024. DiffMM: Multi-Modal Diffusion Model for Recommendation. InProceedings of the 32nd ACM International Conference on Multimedia (MM ’24). ACM, 7591–7599

    2024

  14. [22]

    Haole Ke, Lin Li, PeiPei Wang, Jingling Yuan, and Xiaohui Tao. 2023. Hyperbolic Mutual Learning for Bundle Recommendation. InDatabase Systems for Advanced Applications (DASFAA ’23). Springer Nature Switzerland, 417–433

    2023

  15. [23]

    Jinri Kim, Eungi Kim, Kwangeun Yeo, Yujin Jeon, Chanwoo Kim, Sewon Lee, and Joonseok Lee. 2024. Content-Based Graph Reconstruction for Cold-Start Item Recommendation. InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrie...

    2024

  16. [24]

    Kingma and Max Welling

    Diederik P. Kingma and Max Welling. 2014. Auto-Encoding Variational Bayes. In2nd International Conference on Learning Representations (ICLR ’14)

    2014

  17. [25]

    Miao Li, Xuguang Bao, Liang Chang, Zhoubo Xu, and Long Li. 2020. A Survey of Researches on Personalized Bundle Recommendation Techniques. InMachine Learning for Cyber Security. Springer International Publishing, 290–304

    2020

  18. [26]

    Ming Li, Lin Li, Xiaohui Tao, and Jimmy Xiangji Huang. 2024. MealRec+ : A Meal Recommendation Dataset with Meal- Course Affiliation for Personalization and Healthiness. InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retr...

    2024

  19. [27]

    Ming Li, Lin Li, Xiaohui Tao, Qing Xie, and Jingling Yuan. 2023. Category-Wise Meal Recommendation. InNeural Information Processing - 30th International Conference (ICONIP ’23). Springer, 282–294

    2023

  20. [28]

    Ming Li, Lin Li, Xiaohui Tao, Zhongwei Xie, Qing Xie, and Jingling Yuan. 2024. Boosting Healthiness Exposure in Category-Constrained Meal Recommendation Using Nutritional Standards. 15, 4 (2024), 1–28

    2024

  21. [29]

    Wuchao Li, Rui Huang, Haijun Zhao, Chi Liu, Kai Zheng, Qi Liu, Na Mou, Guorui Zhou, Defu Lian, Yang Song, Wentian Bao, Enyun Yu, and Wenwu Ou. 2025. DimeRec: A Unified Framework for Enhanced Sequential Recommendation via Generative Diffusion Models. InProceedings of the Eighte...

    2025

  22. [30]

    Hashimoto

    Xiang Lisa Li, John Thickstun, Ishaan Gulrajani, Percy Liang, and Tatsunori B. Hashimoto. 2022. Diffusion-LM Improves Controllable Text Generation. InProceedings of the 36th International Conference on Neural Information Processing Systems (NeurIPS ’22). Curran Associates Inc....

    2022

  23. [31]

    2024.A Survey on Diffusion Models for Recommender Systems

    Jianghao Lin, Jiaqi Liu, Jiachen Zhu, Yunjia Xi, Chengkai Liu, Yangtian Zhang, Yong Yu, and Weinan Zhang. 2024.A Survey on Diffusion Models for Recommender Systems. arXiv:2409.05033 [cs]

    2024 arXiv

  24. [32]

    Lakshmanan

    Yidan Liu, Min Xie, and Laks V.S. Lakshmanan. 2014. Recommending User Generated Item Lists. InProceedings of the 8th ACM Conference on Recommender Systems (RecSys ’14). ACM, 185–192

    2014

  25. [33]

    Cheng Lu, Yuhao Zhou, Fan Bao, Jianfei Chen, Chongxuan Li, and Jun Zhu. 2022. DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Steps. InAdvances in Neural Information Processing Systems 35: Annual Conference on Neural Information Processing...

    2022

  26. [34]

    Haokai Ma, Ruobing Xie, Lei Meng, Xin Chen, Xu Zhang, Leyu Lin, and Zhanhui Kang. 2024. Plug-In Diffusion Model for Sequential Recommendation. 38, 8 (2024), 8886–8894. Issue 8

    2024

  27. [35]

    Yunshan Ma, Yingzhi He, Xiang Wang, Yinwei Wei, Xiaoyu Du, Yuyangzi Fu, and Tat-Seng Chua. 2024. MultiCBR: Multi-View Contrastive Learning for Bundle Recommendation. 42, 4 (2024), 100:1–100:23

    2024

  28. [36]

    Yunshan Ma, Yingzhi He, An Zhang, Xiang Wang, and Tat-Seng Chua. 2022. CrossCBR: Cross-View Contrastive Learning for Bundle Recommendation. InProceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’22). ACM, 1233–1241

    2022

  29. [37]

    Alexander Quinn Nichol and Prafulla Dhariwal. 2021. Improved Denoising Diffusion Probabilistic Models. InProceedings of the 38th International Conference on Machine Learning (ICML ’21). PMLR, 8162–8171

    2021

  30. [38]

    Apurva Pathak, Kshitiz Gupta, and Julian McAuley. 2017. Generating and Personalizing Bundle Recommendations on Steam. InProceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR ’17). ACM, 1073–1076

    2017

  31. [39]

    Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme. 2009. BPR: Bayesian Personalized Ranking from Implicit Feedback. InProceedings of the Twenty-Fifth Conference on Uncertainty in Artificial Intelligence (UAI ’09). AUAI Press, 452–461

    2009

  32. [40]

    Ludan Ruan, Yiyang Ma, Huan Yang, Huiguo He, Bei Liu, Jianlong Fu, Nicholas Jing Yuan, Qin Jin, and Baining Guo

  33. [41]

    Shaoyun Shi, Min Zhang, Xinxing Yu, Yongfeng Zhang, Bin Hao, Yiqun Liu, and Shaoping Ma. 2019. Adaptive Feature Sampling for Recommendation with Missing Content Feature Values. InProceedings of the 28th ACM International Conference on Information and Knowledge Management (CIKM...

    2019

  34. [42]

    Harald Steck. 2013. Evaluation of Recommendations: Rating-Prediction and Ranking. InProceedings of the 7th ACM Conference on Recommender Systems (RecSys ’13). ACM, 213–220

    2013

  35. [43]

    Changfeng Sun, Han Liu, Meng Liu, Zhaochun Ren, Tian Gan, and Liqiang Nie. 2020. LARA: Attribute-to-Feature Adversarial Learning for New-Item Recommendation. InProceedings of the 13th International Conference on Web Search and Data Mining (WSDM ’20). ACM, 582–590

    2020

  36. [44]

    2024.A Survey on Bundle Recommendation: Methods, Applications, and Challenges

    Meng Sun, Lin Li, Ming Li, Xiaohui Tao, Dong Zhang, Peipei Wang, and Jimmy Xiangji Huang. 2024.A Survey on Bundle Recommendation: Methods, Applications, and Challenges. arXiv:2411.00341 [cs]

    2024

  37. [45]

    Zhu Sun, Kaidong Feng, Jie Yang, Hui Fang, Xinghua Qu, Yew-Soon Ong, and Wenyuan Liu. 2024. Revisiting Bundle Recommendation for Intent-aware Product Bundling.Trans. Recomm. Syst.2, 3 (2024), 24:1–24:34

    2024

  38. [46]

    Zhu Sun, Jie Yang, Kaidong Feng, Hui Fang, Xinghua Qu, and Yew Soon Ong. 2022. Revisiting Bundle Recommenda- tion: Datasets, Tasks, Challenges and Opportunities for Intent-Aware Product Bundling. InProceedings of the 45th International ACM SIGIR Conference on Research and Deve...

    2022

  39. [47]

    Qian Tao, Chenghao Liu, Yuhan Xia, Yong Xu, and Lusi Li. 2025. Adaptive multi-graph contrastive learning for bundle recommendation.Neural Networks181 (2025), 106832

    2025

  40. [48]

    Maksims Volkovs, Guang Wei Yu, and Tomi Poutanen. 2017. DropoutNet: Addressing Cold Start in Recommender Systems. InAdvances in Neural Information Processing Systems 30: Annual Conference on Neural Information Processing Systems (NeurIPS ’17). 4957–4966

    2017

  41. [49]

    Wenjie Wang, Xinyu Lin, Liuhui Wang, Fuli Feng, Yinwei Wei, and Tat-Seng Chua. 2023. Equivariant Learning for Out-of-Distribution Cold-Start Recommendation. InProceedings of the 31st ACM International Conference on Multimedia (MM ’23). ACM, 903–914

    2023

  42. [50]

    2025.Diffusion Models in Recommendation Systems: A Survey

    Ting-Ruen Wei and Yi Fang. 2025.Diffusion Models in Recommendation Systems: A Survey. arXiv:2501.10548 [cs]

    2025

  43. [51]

    Yinwei Wei, Xiang Wang, Qi Li, Liqiang Nie, Yan Li, Xuanping Li, and Tat-Seng Chua. 2021. Contrastive Learning for Cold-Start Recommendation. InProceedings of the 29th ACM International Conference on Multimedia (MM ’21). ACM, 5382–5390

    2021

  44. [52]

    Sheng, and Guanfeng Liu

    Chuanjiu Wu, Huanhuan Yuan, Pengpeng Zhao, Jianfeng Qu, Victor S. Sheng, and Guanfeng Liu. 2024. Dual-Supervised Contrastive Learning for Bundle Recommendation.IEEE Trans. Comput. Soc. Syst.11, 3 (2024), 3955–3965

    2024

  45. [53]

    Zhenchao Wu and Xiao Zhou. 2023. M2EU: Meta Learning for Cold-Start Recommendation via Enhancing User Preference Estimation. InProceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR ’23). ACM, 1158–1167

    2023

  46. [54]

    Jieyu Yang, Zhaoxin Huan, Yong He, Ke Ding, Liang Zhang, Xiaolu Zhang, Jun Zhou, and Linjian Mo. 2022. Task Similarity Aware Meta Learning for Cold-Start Recommendation. InProceedings of the 31st ACM International Conference on Information & Knowledge Management (CIKM ’22). AC...

    2022

  47. [55]

    Ling Yang, Zhilong Zhang, Yang Song, Shenda Hong, Runsheng Xu, Yue Zhao, Wentao Zhang, Bin Cui, and Ming-Hsuan Yang. 2023. Diffusion Models: A Comprehensive Survey of Methods and Applications.ACM Comput. Surv.56, 4 (2023), 105:1–105:39

    2023

  48. [56]

    Runsheng Yu, Yu Gong, Xu He, Yu Zhu, Qingwen Liu, Wenwu Ou, and Bo An. 2021. Personalized Adaptive Meta Learning for Cold-start User Preference Prediction. InThirty-Fifth AAAI Conference on Artificial Intelligence (AAAI ’21). AAAI Press, 10772–10780

    2021

  49. [57]

    Fuzheng Zhang, Nicholas Jing Yuan, Defu Lian, Xing Xie, and Wei-Ying Ma. 2016. Collaborative Knowledge Base Embedding for Recommender Systems. InProceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD ’16). ACM, 353–362

    2016

  50. [58]

    Dauphin, and David Lopez-Paz

    Hongyi Zhang, Moustapha Cissé, Yann N. Dauphin, and David Lopez-Paz. 2018. mixup: Beyond Empirical Risk Minimization. In6th International Conference on Learning Representations (ICLR ’18). OpenReview.net

    2018

  51. [59]

    Weizhi Zhang, Yuanchen Bei, Liangwei Yang, Henry Peng Zou, Peilin Zhou, Aiwei Liu, Yinghui Li, Hao Chen, Jianling Wang, Yu Wang, Feiran Huang, Sheng Zhou, Jiajun Bu, Allen Lin, James Caverlee, Fakhri Karray, Irwin King, and Philip S. Yu. 2025.Cold-Start Recommendation towards ...

    2025 arXiv

  52. [60]

    Sen Zhao, Wei Wei, Ding Zou, and Xianling Mao. 2022. Multi-View Intent Disentangle Graph Networks for Bundle Recommendation. InThirty-Sixth AAAI Conference on Artificial Intelligence (AAAI ’22). AAAI Press, 4379–4387

    2022

  53. [61]

    Xu Zhao, Yi Ren, Ying Du, Shenzheng Zhang, and Nian Wang. 2022. Improving Item Cold-Start Recommendation via Model-Agnostic Conditional Variational Autoencoder. InProceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval (S...

    2022

  54. [62]

    Zhihui Zhou, Lilin Zhang, and Ning Yang. 2023. Contrastive Collaborative Filtering for Cold-Start Item Recommendation. InProceedings of the ACM Web Conference 2023 (WWW ’23’). ACM, 928–937

    2023

  55. [63]

    Tao Zhu, Patrick Harrington, Junjun Li, and Lei Tang. 2014. Bundle Recommendation in Ecommerce. InProceedings of the 37th International ACM SIGIR Conference on Research & Development in Information Retrieval (SIGIR ’14). ACM, 657–666

    2014

  56. [64]

    Ziwei Zhu, Shahin Sefati, Parsa Saadatpanah, and James Caverlee. 2020. Recommendation for New Users and New Items via Randomized Training and Mixture-of-Experts Transformation. InProceedings of the 43rd International ACM SIGIR Conference on Research and Development in Informat...

    2020

  57. [2023]

    InProceedings of J

    MM-Diffusion: Learning Multi-Modal Diffusion Models for Joint Audio and Video Generation. InProceedings of J. ACM, Vol. 1, No. 1, Article 111. Publication date: January 2025. Divide-and-Conquer: Cold-Start Bundle Recommendation via Mixture of Diffusion Experts 111:27 the IEEE/...

    2025

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.