Pith. sign in

REVIEW 3 major objections 5 minor 5 cited by

EGA-V1: Unifying Online Advertising with End-to-End Learning

T0 review · 3 major / 5 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 A single end-to-end generative model, EGA-V1, can replace the multi-stage cascade architecture for online advertising ranking, producing the final ad sequence directly from the full city-level candidate pool and beating the cascade on…

desk verdict Promising industrial architecture with a genuine online A/B, but the offline evaluation is circular and the complexity analysis has a mismatch—engage, but with caution. read the letter →

arxiv 2505.19755 v2 pith:YKQ4YUPJ submitted 2025-05-26 cs.IR

Junyan Qiu , Ze Wang , Fan Zhang , Zuowu Zheng , Jile Zhu , Jiangke Fan , Teng Zhang , Haitao Wang
show 2 more authors
Yongkang Wang Xingxing Wang
This is my paper · ORCID
classification cs.IR
keywords onlineadvertisingend-to-endlearningmulti-stagecascadearchitecturegenerativerankingnon-autoregressivegenerationclusterattentionreinforcementfromauctionfeedbackaddesign
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 argues that the entire online-advertising ranking pipeline—recall, pre-ranking, ranking, and auction—can be collapsed into one model that generates the final ad sequence directly from the full candidate set in location-based services. It reports that this single-model architecture lifts click-through rate by 5.2%, revenue per mille by 13.6%, and advertiser ROI by 3.1% in online A/B tests against a fully deployed multi-stage cascade, with only a 2.2% relative increase in response time. Offline, it reports eRPM of 217.1 against 194.9 for the strongest cascade baseline, an 11.4% gain, and an incentive-compatibility regret of 2.3% versus 9.1%. If true, the presumed trade-off between computational cost and holistic optimization in advertising disappears: a unified model can account for interactions among all candidate ads instead of only those that survive earlier filtering stages.

What carries the argument

The central object is the cluster-attention block, which replaces the $N \times N$ attention matrix with a learnable cluster matrix $\mathbf{S} \in \mathbb{R}^{N \times N_c}$ that groups keys and values into $N_c$ surrogate tokens, cutting attention cost from $O(N^2 d)$ to $O(N N_c d)$. Global Cluster-Former applies this mechanism within the candidate-ad set and within the user behavior sequence, while Mid-fusion Interest-Former applies it across the two sequences through target attention and context attention, capturing both user interests and ad externalities. AucFormer then maps the resulting representations to slots: slot tokens $\mathbf{T}$ produce an allocation matrix $\mathbf{A} = \mathbf{H}_{\text{ad}} \mathbf{T}^{\top}$, a bid-weighted softmax turns it into allocation probabilities, and a permutation-aware evaluator predicts order-dependent pCTR that also serves as the reward model in reinforcement learning. A payment network, trained with a Lagrangian dual objective, sets payments below bids and approximately satisfies incentive compatibility. The Hybrid Feature Service feeds the whole system by storing ad embeddings locally and user features remotely, so feature retrieval does not become the bottleneck.

What would settle it

Train an independent permutation-aware CTR evaluator on logged exposures from the deployed cascade only, with no EGA-V1 representations or RLAF labels, and re-run the offline eCTR/eRPM comparison; if EGA-V1's advantage shrinks to zero while the online A/B gains also fail to reproduce, the central unification claim is refuted.

Watch

Extended reading notes

Core claim

On the paper's own terms, EGA-V1 establishes that a single non-autoregressive generative model can perform the whole advertising ranking-and-auction task. The model narrows the candidate universe to ads in the same city, roughly $10^5$ candidates per request, and directly outputs the winning sequence for $K$ slots through a bid-weighted allocation matrix, bypassing the stage-by-stage filtering of multi-stage cascades. The reported offline numbers are Recall@50 0.513 versus 0.426 for FS-LTR, AUC 0.754 versus 0.743, eCTR 6.652% versus 6.140%, eRPM 217.1 versus 194.9, and an incentive-compatibility regret of 2.3% versus 9.1%. Online, the paper reports CTR +5.2%, RPM +13.6%, and ROI +3.1% against the deployed cascade. The core assertion is that externality modeling across the full candidate set, enabled by cluster attention and non-autoregressive generation, is what produces these gains.

Load-bearing premise

The load-bearing premise is that the offline replay evaluator—a permutation-aware CTR model trained on EGA-V1's own representations and reused as the RL reward model—scores EGA-V1's output and the cascade baselines without favoring EGA-V1.

Editorial extensions

If this is right

  • If the results hold, advertising platforms can model externalities across the full city-level candidate pool instead of only within a filtered subset, since the same model sees all candidates at once.
  • The non-autoregressive design keeps latency manageable: processing hundreds of times more candidates than the cascade's ranking model costs only 2.2% more response time in the paper's online measurements.
  • The bi-stage training recipe—pre-train on user engagement, then reinforce on auction feedback—lets the generator optimize revenue and advertiser utility directly, and the paper's offline IC metric drops from 9.1% to 2.3%.
  • Cluster attention compensates for the removal of explicit cross-features, so feature storage and transmission can be simplified with much smaller accuracy loss than in cascade models.
  • Scaling the model by stacking cluster-attention blocks gives steady gains with diminishing returns, so the architecture supports future scale-up rather than requiring a pipeline redesign.

Reading between the lines

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

  • Because the offline replay evaluator is trained on EGA-V1's own representations and doubles as the RL reward model, its eCTR and eRPM scores may be biased toward sequences EGA-V1 generates; the online A/B metrics are the independent evidence that would settle the question.
  • The same cluster-attention-plus-NAR-generator-plus-permutation-aware-evaluator recipe could transfer to other large-candidate selection tasks with strong item interactions, such as job search, ride-hailing dispatch, or organic feed ranking, wherever the candidate set can be narrowed by geography or session context.
  • The city-level narrowing to roughly $10^5$ candidates is what makes full-corpus externality modeling tractable; in non-LBS settings without a comparable locality constraint, the unification would need an additional retrieval mechanism or a redefined candidate universe.
  • The 2.3% IC regret is measured under scaled-bid perturbations; testing with adversarial or strategically correlated bid distributions would reveal whether the approximate incentive compatibility is structural or an artifact of the test distribution.
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 / 5 minor

Summary. The paper proposes EGA-V1, an end-to-end generative architecture that replaces a multi-stage cascaded advertising pipeline (recall, pre-ranking, ranking, auction) with a single model. The main components are a Hybrid Feature Service that decouples user and ad features, a RecFormer with cluster-attention for modeling user interest and ad externalities, and an AucFormer with a non-autoregressive generator, a permutation-aware evaluator, and a payment network. Training is done in two stages: pre-training on click/purchase signals, and post-training with reinforcement learning from auction feedback plus a payment-network loss. The paper claims offline improvements in eCTR, eRPM, recall, AUC, and incentive compatibility over MCA and FS-LTR baselines, and online gains of +5.2% CTR, +13.6% RPM, and +3.1% ROI in a one-week A/B test on a low-traffic slot.

Significance. If the evaluation were trustworthy, this would be a meaningful industrial contribution: a single non-autoregressive model that handles both ranking and allocation over a large candidate set, with a concrete system design for feature serving and latency control. The paper is strong in architectural detail and in making the externality and auction-consistency problems explicit. Its main weakness is that the offline evidence is generated by the proposed model's own evaluator, which is also the RLAF reward model, so the headline offline gains may be partially self-referential. The online A/B test is independent but is not accompanied by significance testing and is limited to one week on a low-traffic slot. The complexity analysis also does not reproduce the claimed 0.97 FLOPs ratio. These are load-bearing issues for the central claim that one model outperforms the full cascade pipeline.

major comments (3)
  1. [5.1.2, Eq. (18), Eqs. (21)-(23)] The offline metrics eCTR and eRPM are computed using q_ctr from the permutation-aware evaluator of Eq. (18). That evaluator is trained with Eq. (21), is frozen and used as the reward model in the RLAF objective of Eqs. (22)-(23), and the generator is explicitly optimized to maximize scores produced by it. Therefore the offline comparison of EGA-V1 against MCA and FS-LTR in Table 1 is self-referential: the baselines are scored out-of-distribution by a model whose score function the proposed generator has been trained to exploit. No calibration check, no unbiasedness argument, and no ground-truth click labels are reported for the eCTR/eRPM values. The authors should either validate the evaluator against actual click labels for arbitrary sequences, report offline metrics computed with real user feedback, or provide a separate held-out evaluation that does not use the reward model as the judge.
  2. [5.6, Table 4] The online A/B test is the only independent evidence for the architecture-level claim, but Table 4 reports only relative changes (+5.2% CTR, +13.6% RPM, +3.1% ROI) with no confidence intervals, p-values, sample sizes, or traffic volume for the low-traffic slot. The abstract states that the improvements are 'statistically significant,' yet no significance test is described. Without error bars or a significance statement, the one-week A/B result cannot by itself support the strong conclusion that EGA-V1 outperforms the deployed MCA pipeline. The authors should add significance tests, confidence intervals, and traffic/impression counts, or soften the claim accordingly.
  3. [Appendix A.1] The claimed FLOPs ratio of 0.97 is not reproduced by the paper's own formulas. With the hyperparameters from Section 5.1.4 (N=10^5, L=1000, d=128, N_c=128, m=6, m_k=2, alpha=0.033, N_r=3300), the simplified ratio (2 m_k N L d^2)/(m_r N_r 4 L^2 d) evaluates to approximately 0.65, not 0.97; the full expressions in Eqs. (30)-(33) do not yield 0.97 either. In addition, some printed terms in Eq. (32) appear dimensionally inconsistent as written (e.g., N_ad^2 without a d factor). The authors should correct the formulas, state the exact assumptions used for the ratio, and provide a reproducible per-module FLOPs table.
minor comments (5)
  1. [Abstract and Section 5.1.1] The abstract and conclusion claim 'extensive offline evaluations on public benchmarks,' but Section 5.1.1 describes only the proprietary Meituan dataset; no public benchmark is used anywhere in the experiments. This should be corrected or public experiments should be added.
  2. [Section 5.3, Table 2] The text says 'Judging from the online experimental results in Table 2,' but Table 2 reports offline eCTR and eRPM. The wording should be changed to avoid implying that the ablation results are from online A/B tests.
  3. [Section 5.5] The paragraph before Figure 5 is incomplete: 'Besides, we Results in Figure 5 reveal:' is not a finished sentence, and the figure itself is not described with axis labels or a legend. Please complete the text and make the figure self-contained.
  4. [Section 3.3, footnote 3] The definition of m_k is unclear as printed: the footnote appears to define m_k = ceil(m^2/m_c), which is inconsistent with the hyperparameter values m=6, m_c=2, m_k=2 in Section 5.1.4. Please clarify the intended formula and notation.
  5. [General] There are several small editorial issues: 'calcalated' in Appendix A.1, 'eletment-wise' near Eq. (14), and the notation in Eq. (17) where the softmax is written componentwise as z_{i;k} = Softmax([...])_i but the indexing is confusing. Please proofread and standardize the notation.

Circularity Check

1 steps flagged · score 7.0 of 10

Offline eCTR/eRPM lifts are computed with EGA-V1's own permutation-aware evaluator, the same model used as the RLAF reward, so the +8.3%/+11.4% offline gains are self-referential; only the online A/B is independent.

  1. fitted input called prediction [Section 4.2 (Training the Reward Model, Eq. 21-23) and Section 5.1.2 (Evaluation Metrics, with Eq. 18); Table 1]
    "We further employ an offline replay system with the permutation-aware evaluator to evaluate the expected CTR and the expected revenue of the results generated by different architectures. ... The permutation-aware evaluator in AucFormer serves as a reward model ... Following reward model convergence, the non-autoregressive generator undergoes optimization via Reinforcement Learning from Auction Feedback (RLAF) with frozen evaluator parameters."

    The offline eCTR and eRPM metrics in Section 5.1.2 are defined as sums of q_ctr_yi, the output of the permutation-aware evaluator from Eq. 18. That same evaluator is the RLAF reward model: Eq. 21 trains it on EGA-V1's representations, Eq. 22 computes rewards using its q_ctr_yi, and Eq. 23 maximizes the generator's probability against those rewards. EGA-V1 is therefore explicitly optimized to inflate the very q_ctr_yi scores that later constitute the offline 'ground truth' eCTR/eRPM for EGA-V1, while the MCA and FS-LTR baselines are scored by this EGA-V1-trained evaluator with no reported calibration or unbiasedness check.

full rationale

The principal circular step is the offline evaluation loop: the permutation-aware evaluator (Eq. 18) is trained as part of EGA-V1 (Eq. 21), used as the reward model to optimize the generator (Eqs. 22-23), and then reused as the measurement instrument for eCTR and eRPM in Table 1. The generator is trained to maximize the evaluator's scores, so offline lifts over baselines are at least partly an artifact of optimizing the evaluation metric itself. No ground-truth click labels appear in eCTR/eRPM, and no calibration or unbiasedness validation for the evaluator on out-of-distribution baseline sequences is reported. This is a fitted-input-called-prediction pattern and is the dominant source of the paper's circularity. The online A/B test (Section 5.6) is genuinely independent evidence and supports the architecture-level claim, but it ran for one week on a low-traffic slot, so it does not fully rescue the offline numbers. No load-bearing self-citation chain was found: references to the authors' prior work are contextual, not used to justify the central architectural claim. Separately, the Appendix A.1 FLOPs ratio of 0.97 is not reproduced by the paper's own formulas (plugging N=100,000, N_r=3,300, L=1,000, d=128, m_k=2, m_r=6 into the stated approximation yields about 0.65, not 0.97); this is a correctness concern but not a circularity. The abstract's mention of 'public benchmarks' is unsupported by the experiments, which use only the industrial Meituan dataset, but that too is a reporting issue rather than circular reasoning. Overall, the offline evaluation loop makes the headline offline gains substantially self-referential, warranting a score of 7 despite the independent online A/B evidence.

Assumptions & free parameters 4 free parameters · 5 assumptions · 4 invented entities

The central claim rests on several learned hyperparameters, a self-referential reward model, and an unproven assumption that cluster attention preserves ranking-critical externalities. No independent falsifiable handle exists for the newly proposed system components.

free parameters (4)
  • N_c (number of clusters) = 128
    Set to 128; controls complexity reduction in cluster attention (Section 5.1.4).
  • m, m_e, m_k, N_h, d, L = 6, 3, 2, 4, 128, 1000
    Hyperparameters chosen by grid search (Section 5.1.4).
  • w_z (bid bias weight) = learned, constrained positive
    Learned parameter in allocation probability Eq. 17; forces bid monotonicity.
  • Lagrange multipliers λ_yi and ρ = learned/hyperparameter
    Used in payment network loss Eq. 24 to enforce IC constraints.
assumptions (5)
  • domain assumption The platform objective is to maximize expected sum of payment times pCTR.
    Equation (1) defines revenue as p_i times pCTR_i; this assumes payment and predicted CTR capture true revenue.
  • domain assumption IC and IR are required economic constraints for the auction.
    Standard auction theory from [11]; used as constraints in Equations (3) and (4).
  • ad hoc to paper Cluster attention (Eqs. 7-14) preserves the essential externality information of full attention.
    The paper claims this but provides no theoretical guarantee; only ablation evidence on proprietary data.
  • ad hoc to paper The permutation-aware evaluator's predicted q_ctr is an unbiased estimator of true click probability.
    Used as ground truth in offline replay (eCTR/eRPM) and as reward model in RLAF; if biased, offline results are circular.
  • domain assumption The empirical ex-post regret with M sampled valuations approximates the true IC violation.
    Appendix A.2, Eq. 35, samples M valuations; assumes this estimates IC.
invented entities (4)
  • Hybrid Feature Service (HFS)
    purpose: Decouple user and ad feature processing, storing ad embeddings locally to reduce latency.
    Internal system component; no external falsifiable prediction.
  • Global Cluster-Former (GCF) with adaptive cluster matrix S
    purpose: Reduce attention complexity from O(N^2) to O(N N_c) while modeling ad/ad and user/behavior interactions.
    Internal model component; performance only validated on proprietary data.
  • Mid-fusion Interest-Former (MIF) with context attention
    purpose: Model cross-sequence mutual information between user behavior and candidate ads.
    Internal model component; no external evidence.
  • AucFormer with NAR generator and payment network
    purpose: Generate ad sequences non-autoregressively and optimize payments under auction constraints.
    Internal model component; no external evidence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of EGA-V1: Unifying Online Advertising with End-to-End Learning." pith.science (2026). https://pith.science/paper/YKQ4YUPJ

@misc{pith2026250519755,
  author       = {Pith},
  title        = {Pith review of: EGA-V1: Unifying Online Advertising with End-to-End Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YKQ4YUPJ}},
  note         = {Machine review of arXiv:2505.19755}
}
read the original abstract

Modern industrial advertising systems commonly employ Multi-stage Cascading Architectures (MCA) to balance computational efficiency with ranking accuracy. However, this approach presents two fundamental challenges: (1) performance inconsistencies arising from divergent optimization targets and capability differences between stages, and (2) failure to account for advertisement externalities - the complex interactions between candidate ads during ranking. These limitations ultimately compromise system effectiveness and reduce platform profitability. In this paper, we present EGA-V1, an end-to-end generative architecture that unifies online advertising ranking as one model. EGA-V1 replaces cascaded stages with a single model to directly generate optimal ad sequences from the full candidate ad corpus in location-based services (LBS). The primary challenges associated with this approach stem from high costs of feature processing and computational bottlenecks in modeling externalities of large-scale candidate pools. To address these challenges, EGA-V1 introduces an algorithm and engine co-designed hybrid feature service to decouple user and ad feature processing, reducing latency while preserving expressiveness. To efficiently extract intra- and cross-sequence mutual information, we propose RecFormer with an innovative cluster-attention mechanism as its core architectural component. Furthermore, we propose a bi-stage training strategy that integrates pre-training with reinforcement learning-based post-training to meet sophisticated platform and advertising objectives. Extensive offline evaluations on public benchmarks and large-scale online A/B testing on industrial advertising platform have demonstrated the superior performance of EGA-V1 over state-of-the-art MCAs.

Figures

Figures reproduced from arXiv: 2505.19755 by the authors.

Figure 1
Figure 1. Illustrations of multi-stage cascading architecture [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Architecture Overview of EGA-V1, showcasing its key components including the Hybrid Feature Service, RecFormer [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Architectures of two feature services. prohibitive storage and transmission costs when processing all can￾didate ads per request. This is also a primary reason why advertising systems are constrained to adopt MCA for gradual filtering. Second, repeated RPC communications between feature services and mul￾tiple stages incur substantial latency overhead, particularly when handling high-frequency requests in real-time a… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Three paradigms for user interest modeling. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Scaling trends featuring the relationship between [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 5 Pith papers

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

  1. HiGR: Industrial-Scale Hierarchical Generative Slate Recommendation Framework in Tencent

    cs.IR 2025-12 conditional novelty 7.0 of 10

    A hierarchical generative slate recommender that plans list-level preferences before decoding items achieves 5× speedup and gains in deployed A/B tests.

  2. Gryphon-v2: One Model in Place of a Cascade - Generate-and-Rank Recommender with Rollout Distillation

    cs.IR 2026-08 conditional novelty 6.0 of 10

    A single generate-and-rank model with a distilled ranking module replaces a 15-plus stage production recommender cascade and lifts active users by 1.41% in an online A/B test.

  3. OneSug: The Unified End-to-End Generative Framework for E-commerce Query Suggestion

    cs.IR 2025-06 conditional novelty 6.0 of 10

    A unified encoder-decoder model with prefix representation enhancement and reward-weighted DPO outperforms Kuaishou's online multi-stage query suggestion system in offline and live A/B evaluations.

  4. NGA: Non-autoregressive Generative Auction with Global Externalities for Advertising Systems

    cs.IR 2025-06 conditional novelty 5.0 of 10

    NGA is a non-autoregressive generative auction that models effects of adjacent organic content and computes rewards and payments in parallel, reporting gains in RPM, CTR, CVR, and latency over CGA.

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

    cs.IR 2025-07 unverdicted novelty 3.0 of 10

    A survey of LLM-based generative recommendation systems, covering application settings, training pipelines, industrial deployment challenges, and future directions.

Reference graph

Works this paper leans on

40 extracted references · 11 canonical work pages · cited by 5 Pith papers

  1. [1]

    Paul Covington, Jay Adams, and Emre Sargin. 2016. Deep neural networks for youtube recommendations. InProceedings of the 10th ACM conference on recommender systems. 191–198

    2016

  2. [2]

    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 2025

  3. [3]

    Paul Dütting, Zhe Feng, Harikrishna Narasimhan, David Parkes, and Sai Srivatsa Ravindranath. 2019. Optimal auctions through deep learning. InInternational Conference on Machine Learning. PMLR, 1706–1715

    work page 2019

  4. [4]

    Benjamin Edelman, Michael Ostrovsky, and Michael Schwarz. 2007. Internet advertising and the generalized second-price auction: Selling billions of dollars worth of keywords.American economic review97, 1 (2007), 242–259

    2007

  5. [5]

    Siyu Gu and Xiangrong Sheng. 2022. On Ranking Consistency of Pre-ranking Stage.arXiv preprint arXiv:2205.01289(2022)

    arXiv 2022

  6. [6]

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2016. Deep residual learning for image recognition. InProceedings of the IEEE conference on computer vision and pattern recognition. 770–778

    2016

  7. [7]

    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. 173–182

    2017

  8. [8]

    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. InProceedings of the 22nd ACM international conference on Information & Knowledge Management. 2333–2338

    2013

Show all 40 references
  1. [9]

    Bernard J Jansen and Tracy Mullen. 2008. Sponsored search: an overview of the concept, history, and technology.International Journal of Electronic Business6, 2 (2008), 114–131

    2008

  2. [10]

    Wang-Cheng Kang and Julian McAuley. 2018. Self-attentive sequential recom- mendation. In2018 IEEE international conference on data mining (ICDM). IEEE, 197–206

    2018

  3. [11]

    Ningyuan Li, Yunxuan Ma, Yang Zhao, Zhijian Duan, Yurong Chen, Zhilin Zhang, Jian Xu, Bo Zheng, and Xiaotie Deng. 2023. Learning-based ad auction design with externalities: the framework and a matching-based approach. InProceedings of the 29th ACM SIGKDD Conference on Knowledg...

    2023

  4. [12]

    Xiangyang Li, Bo Chen, HuiFeng Guo, Jingjie Li, Chenxu Zhu, Xiang Long, Sujian Li, Yichao Wang, Wei Guo, Longxia Mao, et al. 2022. Inttower: the next generation of two-tower model for pre-ranking system. InProceedings of the 31st ACM International Conference on Information & K...

    2022

  5. [13]

    Xuejian Li, Ze Wang, Bingqi Zhu, Fei He, Yongkang Wang, and Xingxing Wang

  6. [14]

    Guogang Liao, Xiaowen Shi, Ze Wang, Xiaoxu Wu, Chuheng Zhang, Yongkang Wang, Xingxing Wang, and Dong Wang. 2022. Deep page-level interest net- work in reinforcement learning for ads allocation. InProceedings of the 45th International ACM SIGIR Conference on Research and Develo...

    2022

  7. [15]

    Guogang Liao, Ze Wang, Xiaoxu Wu, Xiaowen Shi, Chuheng Zhang, Yongkang Wang, Xingxing Wang, and Dong Wang. 2022. Cross dqn: Cross deep q network for ads allocation in feed. InProceedings of the ACM Web Conference 2022. 401– 409

    2022

  8. [16]

    Qi Liu, Kai Zheng, Rui Huang, Wuchao Li, Kuo Cai, Yuan Chai, Yanan Niu, Yiqun Hui, Bing Han, Na Mou, et al . 2024. RecFlow: An Industrial Full Flow Recommendation Dataset.arXiv preprint arXiv:2410.20868(2024)

    2024 arXiv

  9. [17]

    Xiangyu Liu, Chuan Yu, Zhilin Zhang, Zhenzhe Zheng, Yu Rong, Hongtao Lv, Da Huo, Yiqing Wang, Dagui Chen, Jian Xu, et al. 2021. Neural auction: End-to- end learning of auction mechanisms for e-commerce advertising. InProceedings of the 27th ACM SIGKDD Conference on Knowledge D...

    2021

  10. [18]

    Xinchen Luo, Jiangxia Cao, Tianyu Sun, Jinkai Yu, Rui Huang, Wei Yuan, Hezheng Lin, Yichen Zheng, Shiyao Wang, Qigen Hu, et al . 2024. QARM: Quantita- tive Alignment Multi-Modal Recommendation at Kuaishou.arXiv preprint arXiv:2411.11739(2024)

    2024 arXiv

  11. [19]

    Tomas Mikolov, Ilya Sutskever, Kai Chen, Greg S Corrado, and Jeff Dean. 2013. Distributed representations of words and phrases and their compositionality. Advances in neural information processing systems26 (2013)

    2013

  12. [20]

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. Training language models to follow instructions with human feedback.Advances in neural information processing systems35 (...

    2022

  13. [21]

    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. InProceedings of the 29th ACM International Conference on Informati...

    2020

  14. [22]

    Jiarui Qin, Jiachen Zhu, Bo Chen, Zhirong Liu, Weiwen Liu, Ruiming Tang, Rui Zhang, Yong Yu, and Weinan Zhang. 2022. Rankflow: Joint optimization of multi- stage cascade ranking systems as flows. InProceedings of the 45th International ACM SIGIR Conference on Research and Deve...

    2022

  15. [23]

    Shashank Rajput, Nikhil Mehta, Anima Singh, Raghunandan Hulikal Keshavan, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Tran, Jonah Samost, et al

  16. [24]

    Adjorn Van Engelenhoven, Nicola Strisciuglio, and Estefanía Talavera. 2024. CAST: Clustering Self-Attention using Surrogate Tokens for Efficient Transform- ers.arXiv preprint arXiv:2402.04239(2024)

    2024 arXiv

  17. [25]

    Recommender systems with generative retrieval.Advances in Neural Information Processing Systems36 (2024)

    2024

  18. [26]

    Lidan Wang, Jimmy Lin, and Donald Metzler. 2011. A cascade ranking model for efficient ranked retrieval. InProceedings of the 34th international ACM SIGIR conference on Research and development in Information Retrieval. 105–114

    2011

  19. [27]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need.Advances in neural information processing systems30 (2017)

    2017

  20. [28]

    Zhe Wang, Liqin Zhao, Biye Jiang, Guorui Zhou, Xiaoqiang Zhu, and Kun Gai

  21. [29]

    Yunli Wang, Zhiqiang Wang, Jian Yang, Shiyang Wen, Dongying Kong, Han Li, and Kun Gai. 2024. Adaptive Neural Ranking Framework: Toward Maximized Business Goal for Cascade Ranking Systems. InProceedings of the ACM on Web Conference 2024. 3798–3809

    2024

  22. [30]

    Jiaqi Zhai, Lucy Liao, Xing Liu, Yueming Wang, Rui Li, Xuan Cao, Leon Gao, Zhaojie Gong, Fangda Gu, Jiayuan He, et al . 2024. Actions speak louder than words: trillion-parameter sequential transducers for generative recommendations. InProceedings of the 41st International Conf...

    2024

  23. [31]

    Zhixuan Zhang, Yuheng Huang, Dan Ou, Sen Li, Longbin Li, Qingwen Liu, and Xiaoyi Zeng. 2023. Rethinking the role of pre-ranking in large-scale e-commerce searching system.arXiv preprint arXiv:2305.13647(2023)

    2023 arXiv

  24. [32]

    Yuhao Yang, Zhi Ji, Zhaopeng Li, Yi Li, Zhonglin Mo, Yue Ding, Kai Chen, Zijian Zhang, Jie Li, Shuanglong Li, et al. 2025. Sparse Meets Dense: Unified Generative Recommendations with Cascaded Sparse-Dense Representations.arXiv preprint arXiv:2503.02453(2025)

    2025 arXiv

  25. [33]

    Zhishan Zhao, Jingyue Gao, Yu Zhang, Shuguang Han, Siyuan Lou, Xiang-Rong Sheng, Zhe Wang, Han Zhu, Yuning Jiang, Jian Xu, et al. 2023. COPR: Consistency- Oriented Pre-Ranking for Online Advertising. InProceedings of the 32nd ACM International Conference on Information and Kno...

    2023

  26. [34]

    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. In2024 IEEE 40th International Conference on Data Engineering (ICDE). IEEE, 1435–1448

    2024

  27. [35]

    Binglei Zhao, Houying Qi, Guang Xu, Mian Ma, Xiwei Zhao, Feng Mei, Sulong Xu, and Jinghe Hu. 2025. A Hybrid Cross-Stage Coordination Pre-ranking Model for Online Recommendation Systems.arXiv preprint arXiv:2502.10284(2025)

    2025 arXiv

  28. [36]

    Guorui Zhou, Xiaoqiang Zhu, Chenru Song, Ying Fan, Han Zhu, Xiao Ma, Yanghui Yan, Junqi Jin, Han Li, and Kun Gai. 2018. Deep interest network for click-through rate prediction. InProceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining. ...

    2018

  29. [37]

    Ruitao Zhu, Yangsu Liu, Dagui Chen, Zhenjia Ma, Chufeng Shi, Zhenzhe Zheng, Jie Zhang, Jian Xu, Bo Zheng, and Fan Wu. 2024. Contextual Generative Auction with Permutation-level Externalities for Online Advertising.arXiv preprint arXiv:2412.11544(2024)

    2024 arXiv

  30. [38]

    Kai Zheng, Haijun Zhao, Rui Huang, Beichuan Zhang, Na Mou, Yanan Niu, Yang Song, Hongning Wang, and Kun Gai. 2024. Full stage learning to rank: A unified framework for multi-stage systems. InProceedings of the ACM Web Conference

    2024

  31. [2020]

    Cold: Towards the next generation of pre-ranking system.arXiv preprint arXiv:2007.16122(2020)

    2020 arXiv

  32. [2024]

    InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval

    Deep automated mechanism design for integrating ad auction and allocation in feed. InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. 1211–1220

Pith tools

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