Pith. sign in

REVIEW 3 major objections 4 minor 35 references

Optimal Auction Design in the Joint Advertising

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read For one ad slot, the revenue-maximizing joint-auction mechanism is a step function that awards the slot to the bundle with the highest combined virtual value and charges each winner the critical bid.

desk verdict Plausible and novel single-slot optimality result for joint ads, but the appendix proof has a sign error and the experiments overclaim state-of-the-art; worth a serious referee, not acceptance as-is. read the letter →

arxiv 2507.07418 v1 pith:V6JJJ422 submitted 2025-07-10 cs.GT cs.AI

Yang Li , Yuchao Ma , Qi Qi This is my paper
classification cs.GTcs.AI MSC 91B2691B0368T07
keywords jointadvertisingauctionmechanismdesignrevenuemaximizationvirtualvaluescriticalvaluepaymentsneuralincentivecompatibilityindividualrationality
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 sets out to find the revenue-maximizing auction rule for joint advertising, in which a single ad slot is awarded to a bundle made of one retailer and one supplier who bid together. It claims that in the single-slot case the optimal deterministic mechanism is a step function: award the slot to the bundle whose combined virtual values clear the auctioneer's reserve and beat every competing bundle, and charge each winning bidder the critical bid at which that bundle would just lose. For the multi-slot case, where no characterization is claimed, it introduces BundleNet, a neural network that learns allocation and payment rules from bundle-level regret constraints, and reports that it tracks the theoretical optimum on single-slot problems and raises revenue over earlier mechanisms on multi-slot problems while keeping incentive-compatibility violations near zero. The result matters because joint advertising is a real revenue format on large platforms, and no previous mechanism for it had an optimality guarantee.

What carries the argument

The carrying object is the bundle virtual value, defined as the sum of the two partners' virtual values, $c_e(v_r,v_s)=c_r(v_r)+c_s(v_s)$, where each bidder's virtual value is the standard single-parameter transform $c_i(v)=v-(1-F_i(v))/f_i(v)$. Because the slot goes to the bundle with the highest such sum above the reserve, each bidder only needs to track the neighboring partner with the highest virtual value, which reduces the two-sided joint-bid problem to a one-dimensional step-function allocation with a critical-value payment. In the multi-slot neural mechanism, the carrying device is a bundle-level regret constraint whose sum is shown to upper-bound the sum of per-bidder regret, so driving bundle regrets to zero pushes the whole mechanism toward dominant-strategy incentive compatibility.

What would settle it

Take a single-slot joint-ad environment whose values are sampled from a distribution with non-monotone virtual values or correlated retailer and supplier values, compute the expected revenue of the step-function critical-payment mechanism, and compare it with an unrestricted mechanism on the same draw distribution; if the unrestricted mechanism reliably earns more, or if truthful bidding is not a dominant strategy under the step-function rule, the optimality claim is false.

Watch

Extended reading notes

Core claim

The central discovery is a necessary-and-sufficient characterization of optimal joint auctions with a single slot (Theorem 4.3). Under regular value distributions, a deterministic joint auction maximizes expected revenue if and only if its allocation is a step function that gives the slot to the bundle with the highest sum of the two partners' virtual values, provided that sum exceeds the auctioneer's reserve value $v_0$, and its payments charge each winner the critical value, defined as the lowest bid at which the winner's best bundle would still clear the reserve and beat all bundles that exclude the winner. The experiments then show that BundleNet's learned mechanism approximates this step-function allocation in the single-slot setting and delivers the reported revenue improvements in the multi-slot setting against prior learned and incentive-compatible baselines.

Load-bearing premise

The theorem assumes each advertiser's value is drawn independently from a known regular distribution and that a bundle's value is exactly the sum of the two partners' values, so if real joint-ad values are correlated, non-additive, or non-regular, the step-function rule need not be revenue-optimal.

Editorial extensions

If this is right

  • A single-slot joint-ad auction has an exactly characterized revenue-optimal rule, so platforms can test any deterministic mechanism they run against the step-function and critical-value prescription.
  • Winning payments have a clean operational meaning: each winning partner pays the lowest bid that keeps their bundle ahead of the auctioneer's reserve and all rival bundles.
  • The bundle-level regret constraint gives a reusable substitute for per-bidder incentive constraints in multi-slot joint auctions, because it dominates the per-bidder constraints by Lemma 5.1.
  • The reported experiments imply that encoding the bundle structure explicitly, rather than treating partners as independent bidders, is what lets a learned mechanism approach the theoretical optimum.

Reading between the lines

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

  • Our inference: if the single-slot characterization carries over to multiple slots, the optimum should be a weighted virtual-value rank using the slot click-through rates $\lambda_k$; checking BundleNet's learned allocations against such a weighted rule is a direct test.
  • Our inference: the additive bundle-value assumption is the first point of failure in practice; a testable extension would re-derive the critical values under correlated retailer-supplier values, and the theorem's step-function rule should not be trusted under such correlation.
  • Our inference: the mechanism assumes the auctioneer knows the true distributions, so a practical deployment would need estimated distributions; measuring the revenue loss of the step-function rule under distributional misspecification would quantify how much robustness costs.
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 studies joint advertising auctions, in which an ad slot is allocated to a bundle consisting of a retailer and a supplier rather than to a single advertiser. The main theoretical contribution is Theorem 4.3, an 'if and only if' characterization of the revenue-maximizing mechanism for a single-slot joint advertisement under regularity, modeled as a Myerson-style single-parameter auction with additively combined virtual values. The paper then proposes BundleNet, a neural network architecture for multi-slot joint advertising, which uses bundle-level incentive-compatibility constraints and is evaluated against RVCG, JRegNet, and the single-slot optimal benchmark on synthetic data.

Significance. If Theorem 4.3 were properly established, it would provide a clean, exact Myerson-type solution for single-slot joint advertising, and it would serve as a useful theoretical benchmark for learning-based mechanisms. The BundleNet architecture is a reasonable extension of RegretNet to a bipartite bundle structure, and the idea of formulating IC constraints at the bundle level is motivated by the problem structure. The paper is useful in bringing these two threads together, but the printed proof of the central theorem contains a sign error in the revenue decomposition, and the 'if and only if' direction is not demonstrated. The empirical state-of-the-art claim is also stronger than the results in Table 5 support. The underlying ideas appear sound and repairable, but the manuscript as written does not yet establish its main claims.

major comments (3)
  1. [Appendix B, Eq. (8)] The sign error in Eq. (8) breaks the argument that individual rationality binds at the lowest type. Substituting Eq. (7) into Eq. (6) yields U0 = ∫ v0 f + ∑_e ∫ x_e(v_s + v_r − (1−F_r)/f_r − (1−F_s)/f_s − v0) f dv − ∑_r U_r(a_r,a_r) − ∑_s U_s(a_s,a_s). As printed, Eq. (8) has plus signs before the two U sums. With the printed plus signs, the statement that these non-negative terms should be set to zero to maximize revenue is backwards; with the correct minus signs, the conclusion follows. Since this step is the origin of the payment rule in Eq. (9), Theorem 4.3 is not supported by the printed derivation.
  2. [Appendix B, proof of Theorem 4.3] The proof of the 'if and only if' claim is incomplete. After writing the virtual surplus objective, the text says 'This implicitly indicates that the slot will be allocated to the bundle with the highest virtual value,' but no formal equivalence is proved between the virtual-surplus maximizer and the step-function allocation with critical-value payments stated in Theorem 4.3. The necessity direction is absent, tie cases at v_i = v̂_i are not discussed, and the payment calculation is only summarized as 'we integrate the allocation rule.' In addition, the theorem does not explicitly state a bundle-level allocation rule, so the relation between x_i^M(v_i,v_−i) and the identity of the winning bundle is left informal.
  3. [Table 5 (Appendix D.2)] The abstract and Section 6.2 claim that BundleNet achieves state-of-the-art performance, but Table 5 shows that in the N5×5 setting JRegNet's revenue (2.2071) exceeds BundleNet's (2.1393). The claim should be qualified as 'in most settings' or the state-of-the-art statement should be based on a defined aggregation across configurations. As written, the empirical claim is stronger than the data support.
minor comments (4)
  1. [Definition 3.1] The allocation rule is first defined as xe : V → 2^M, a set-valued mapping, but later xe(v) is treated as a scalar or a vector; this notation should be made consistent.
  2. [Eq. (4) and Algorithm 1] Eq. (4) sums over ℓ = 0 to L while the rest of the paper uses ℓ = 1 to L, and the Lagrange multipliers are denoted inconsistently as λ, µ, and μ across Section 5.3 and Algorithm 1.
  3. [Appendix B, final displayed equations] In the last payment equation of Appendix B, the winner's payment is written as z_r(v_−r) without any definition of z_r; it should be v̂_r(v_−r) as in Theorem 4.3.
  4. [Section 5.2] The doubly stochastic construction uses matrices of size (n+1) × (m+1), but Lemma 5.2 is stated for n × m matrices; the role of the dummy row and column and the treatment of the resulting allocation probabilities should be clarified.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central optimality result derives from Myerson's external theorem and the neural-network comparison is against an independently computed benchmark.

full rationale

The load-bearing theoretical claim (Theorem 4.3) is derived in Appendix B from Myerson's Lemma (Lemma B.1, cited to Myerson 1981), an external characterization, and from a virtual-surplus maximization (Lemma B.2) that does not presuppose the theorem being proved. The critical-value payment rule is the standard Myerson payment identity, not a fitted parameter renamed as a prediction. BundleNet is evaluated by comparing its revenue and regret against this independently computed optimal mechanism; no quantity learned by BundleNet is fed back into the proof of Theorem 4.3. The paper cites the authors' own prior RVCG/JRegNet work, but only as baselines and for the joint-advertising model definition; those citations are not used to justify optimality. The proof as printed has a sign issue in Eq. (8): substituting Eq. (7) into Eq. (6) gives the low-type utility terms with a minus sign, while the displayed Eq. (8) prints them with a plus sign, and the sentence 'This implicitly indicates that the slot will be allocated to the bundle with the highest virtual value' is terse. These are soundness and rigor concerns, not circular reductions, because the argument's source remains Myerson's external characterization rather than the theorem's own conclusion. The experimental comparison against the Optimal baseline is therefore not a self-referential validation, and no circular step is exhibited.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The paper's theory is a direct application of Myerson's single-parameter framework to a two-bidder-per-bundle variant of sponsored search. It introduces no new physical or informational entities. The load-bearing assumptions are the standard independence and regularity conditions, the additive-value model, and the known-distribution assumption; the mechanism's revenue scales with v0, which is fixed to 0 in experiments.

free parameters (2)
  • v0 (auctioneer's value for an unallocated slot) = 0 in all experiments
    The revenue formula and the critical values in Theorem 4.3 depend on v0; the paper sets v0=0 in experiments without sensitivity analysis.
  • Synthetic value distribution parameters = U(0,1), truncated Exp(2), truncated N(0.5,0.1), truncated LN(0.1,1.44)
    Experimental settings are chosen by hand; results may not generalize to other value distributions, and no error bars or sensitivity analysis are provided.
assumptions (5)
  • domain assumption Independent private values: every retailer and supplier value is drawn independently from a known prior distribution F.
    Stated in Section 3 and used throughout; Theorem 4.3's critical values are defined against these distributions.
  • domain assumption Regularity of distributions (virtual values are increasing), Definition 4.2.
    Theorem 4.3 assumes regular bidders to guarantee monotonicity of the allocation; without regularity the virtual-surplus maximizer may violate incentive compatibility.
  • standard math Myerson's Lemma and the feasibility characterization (Lemma B.1, Myerson 1981).
    The optimality proof in Appendix B reduces to Myerson's theory for single-parameter auctions; this is cited, not re-derived.
  • domain assumption Bundle value is the additive sum of the two partners' values, and each participant cares only about whether their own ad is shown.
    Defined in Definition 3.1; if a retailer cares which supplier is matched or values are non-additive, the mechanism is not optimal.
  • domain assumption The bipartite graph G=(R,S,E) of eligible bundles is exogenous and fixed.
    The allocation and critical values in Theorem 4.3 treat the neighbor structure as given; in practice eligible bundles can change with the search query.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Optimal Auction Design in the Joint Advertising." pith.science (2026). https://pith.science/paper/V6JJJ422

@misc{pith2026250707418,
  author       = {Pith},
  title        = {Pith review of: Optimal Auction Design in the Joint Advertising},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/V6JJJ422}},
  note         = {Machine review of arXiv:2507.07418}
}
read the original abstract

Online advertising is a vital revenue source for major internet platforms. Recently, joint advertising, which assigns a bundle of two advertisers in an ad slot instead of allocating a single advertiser, has emerged as an effective method for enhancing allocation efficiency and revenue. However, existing mechanisms for joint advertising fail to realize the optimality, as they tend to focus on individual advertisers and overlook bundle structures. This paper identifies an optimal mechanism for joint advertising in a single-slot setting. For multi-slot joint advertising, we propose \textbf{BundleNet}, a novel bundle-based neural network approach specifically designed for joint advertising. Our extensive experiments demonstrate that the mechanisms generated by \textbf{BundleNet} approximate the theoretical analysis results in the single-slot setting and achieve state-of-the-art performance in the multi-slot setting. This significantly increases platform revenue while ensuring approximate dominant strategy incentive compatibility and individual rationality.

Figures

Figures reproduced from arXiv: 2507.07418 by the authors.

Figure 1
Figure 1. A Comparison Between the Traditional Advertising Model and the Joint Advertising Model.The image on the left illustrates traditional advertising, where the retailer submits a bid for the ad, while the supplier (brand owner) does not participate in the auction. In this case, the platform only receives the bid from the retailer. In contrast, as shown in the right image, for a joint advertisement, both the retailer and… view at source ↗
Figure 2
Figure 2. The neural network architecture BundleNet. Details are shown in Sec 5.2. (CPC), defined as Xr = brλ ∈ R m ≥0 for retailer r or Xs = bsλ ∈ R m ≥0 for supplier s. To capture the combined influence of retailers and suppliers on each bundle, we aggregate their node features into edge features, which we refer to as Divided Bids for the Bundle. This process is mathematically formalized as follows: DBe = [Xr, Xs] ∈ R 2m ≥0… view at source ↗
Figure 3
Figure 3. The figure presents the allocation rules learned by BundleNet and JRegNet under the Setting U2. Subfigures (a) and (c) show the allocation results of BundleNet in two experiments, while (b) and (d) display the results of JRegNet in the same experiments. The solid regions in all subfigures depict the probability of the single-slot being allocated to bundle e1, with the white dashed line representing the boundary of t… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: The figure presents the allocation rules learned by BundleNet and JRegNet under the Setting U2. Subfigure (1) show the allocation results of optimal mechanism, BundleNet and JRegNet in the first experiment, while subfigure (2) display the results of three mechanism in …

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

35 extracted references · 33 canonical work pages

  1. [1]

    write newline

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

  2. [2]

    Truthful auctions for pricing search keywords

    Aggarwal, G., Goel, A., and Motwani, R. Truthful auctions for pricing search keywords. In Proceedings of the 7th ACM Conference on Electronic Commerce, pp.\ 1--7. ACM, 2006

    work page 2006

  3. [3]

    Selling joint ads: A regret minimization perspective

    Aggarwal, G., Badanidiyuru, A., D \"u tting, P., and Fusco, F. Selling joint ads: A regret minimization perspective. In Proceedings of the 25th ACM Conference on Economics and Computation, pp.\ 164--194, 2024

    work page 2024

  4. [4]

    R., and Sivan, B

    Charles, D., Devanur, N. R., and Sivan, B. Multi-score position auctions. In Proceedings of the Ninth ACM International Conference on Web Search and Data Mining, pp.\ 417--425, 2016

    work page 2016

  5. [5]

    Clarke, E. H. Multipart pricing of public goods. Public choice, pp.\ 17--33, 1971

    work page 1971

  6. [6]

    and Sandholm, T

    Conitzer, V. and Sandholm, T. Complexity of mechanism design. In Proceedings of the Eighteenth Conference on Uncertainty in Artificial Intelligence, pp.\ 103–110, 2002. ISBN 1558608974

    work page 2002

  7. [7]

    and Sandholm, T

    Conitzer, V. and Sandholm, T. Self-interested automated mechanism design and implications for optimal combinatorial auctions. In Proceedings of the 5th ACM Conference on Electronic Commerce, pp.\ 132--141, 2004

    work page 2004

  8. [8]

    Certifying strategyproof auction networks

    Curry, M., Chiang, P.-Y., Goldstein, T., and Dickerson, J. Certifying strategyproof auction networks. Advances in Neural Information Processing Systems, 33: 0 4987--4998, 2020

    work page 2020

Show all 35 references
  1. [9]

    Differentiable economics for randomized affine maximizer auctions

    Curry, M., Sandholm, T., and Dickerson, J. Differentiable economics for randomized affine maximizer auctions. arXiv preprint arXiv:2202.02872, 2022

    2022 arXiv

  2. [10]

    A context-integrated transformer-based neural network for auction design

    Duan, Z., Tang, J., Yin, Y., Feng, Z., Yan, X., Zaheer, M., and Deng, X. A context-integrated transformer-based neural network for auction design. In International Conference on Machine Learning, pp.\ 5609--5626. PMLR, 2022

    2022

  3. [11]

    A scalable neural network for dsic affine maximizer auction design

    Duan, Z., Sun, H., Chen, Y., and Deng, X. A scalable neural network for dsic affine maximizer auction design. Advances in Neural Information Processing Systems, 36: 0 56169--56185, 2023

    2023

  4. [12]

    C., and Ravindranath, S

    D \"u tting, P., Feng, Z., Narasimhan, H., Parkes, D. C., and Ravindranath, S. S. Optimal auctions through deep learning: Advances in differentiable economics. Journal of the ACM, 71 0 (1): 0 1--53, 2024

    2024

  5. [13]

    Internet advertising and the generalized second-price auction: Selling billions of dollars worth of keywords

    Edelman, B., Ostrovsky, M., and Schwarz, M. Internet advertising and the generalized second-price auction: Selling billions of dollars worth of keywords. American economic review, 97 0 (1): 0 242--259, 2007

    2007

  6. [14]

    Facebook collaborative ads, 2024

    Facebook. Facebook collaborative ads, 2024. URL https://www.facebook.com/business/tools/collaborative-ads. [Online; accessed 02-February-2024]

    2024

  7. [15]

    Incentives in teams

    Groves, T. Incentives in teams. Econometrica: Journal of the Econometric Society, pp.\ 617--631, 1973

    1973

  8. [16]

    Optimizing affine maximizer auctions via linear programming: an application to revenue maximizing mechanism design for zero-day exploits markets

    Guo, M., Hata, H., and Babar, A. Optimizing affine maximizer auctions via linear programming: an application to revenue maximizing mechanism design for zero-day exploits markets. In PRIMA 2017: Principles and Practice of Multi-Agent Systems: 20th International Conference, pp.\...

    2017

  9. [17]

    Optimal-er auctions through attention

    Ivanov, D., Safiulin, I., Filippov, I., and Balabaeva, K. Optimal-er auctions through attention. Advances in Neural Information Processing Systems, 35: 0 34734--34747, 2022

    2022

  10. [18]

    and Pennock, D

    Lahaie, S. and Pennock, D. M. Revenue analysis of a family of ranking rules for keyword auctions. In Proceedings of the 8th ACM Conference on Electronic Commerce, pp.\ 50--56, 2007

    2007

  11. [19]

    Nma: Neural multi-slot auctions with externalities for online advertising

    Liao, G., Li, X., Wang, Z., Yang, F., et al. Nma: Neural multi-slot auctions with externalities for online advertising. arXiv preprint arXiv:2205.10018, 2022

    2022 arXiv

  12. [20]

    and Sandholm, T

    Likhodedov, A. and Sandholm, T. Approximating revenue-maximizing combinatorial auctions. In Proceedings of the AAAI Conference on Artificial Intelligence. AAAI, 2004

    2004

  13. [21]

    Neural auction: End-to-end learning of auction mechanisms for e-commerce advertising

    Liu, X., Yu, C., Zhang, Z., Zheng, Z., Rong, Y., Lv, H., Huo, D., Wang, Y., Chen, D., Xu, J., et al. Neural auction: End-to-end learning of auction mechanisms for e-commerce advertising. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, pp....

    2021

  14. [22]

    Joint bidding in ad auctions

    Ma, Y., Li, W., Zhang, W., Lei, Y., Zhang, Z., Qi, Q., Liu, Q., and Wang, X. Joint bidding in ad auctions. In Annual Conference on Theory and Applications of Models of Computation, pp.\ 344--354. Springer, 2024

    2024

  15. [23]

    Myerson, R. B. Optimal auction design. Mathematics of operations research, 6 0 (1): 0 58--73, 1981

    1981

  16. [24]

    Preferencenet: Encoding human preferences in auction design with deep learning

    Peri, N., Curry, M., Dooley, S., and Dickerson, J. Preferencenet: Encoding human preferences in auction design with deep learning. Advances in Neural Information Processing Systems, 34: 0 17532--17542, 2021

    2021

  17. [25]

    Rahme, J., Jelassi, S., Bruna, J., and Weinberg, S. M. A permutation-equivariant neural network architecture for auction design. In Proceedings of the AAAI conference on artificial intelligence, volume 35, pp.\ 5664--5672, 2021

    2021

  18. [26]

    A., and Key, P

    Roberts, B., Gunawardena, D., Kash, I. A., and Key, P. Ranking and tradeoffs in sponsored search auctions. ACM Transactions on Economics and Computation, 4 0 (3): 0 1--21, 2016

    2016

  19. [27]

    The characterization of implementable choise rules

    Roberts, K. The characterization of implementable choise rules. In Aggregation and Revelation of Preferences, pp.\ 321--349. North-Holland, 1979

    1979

  20. [28]

    and Likhodedov, A

    Sandholm, T. and Likhodedov, A. Automated design of revenue-maximizing combinatorial auctions. Operations Research, 63 0 (5): 0 1000--1025, 2015

    2015

  21. [29]

    Automated mechanism design via neural networks

    Shen, W., Tang, P., and Zuo, S. Automated mechanism design via neural networks. pp.\ 215–223. International Foundation for Autonomous Agents and Multiagent Systems, 2019. ISBN 9781450363099

    2019

  22. [30]

    Thompson, D. R. and Leyton-Brown, K. Revenue optimization in the generalized second-price auction. In Proceedings of the fourteenth ACM conference on Electronic commerce, pp.\ 837--852, 2013

    2013

  23. [31]

    Varian, H. R. Position auctions. International Journal of Industrial Organization, 25 0 (6): 0 1163--1178, 2007

    2007

  24. [32]

    Counterspeculation, auctions, and competitive sealed tenders

    Vickrey, W. Counterspeculation, auctions, and competitive sealed tenders. The Journal of finance, 16 0 (1): 0 8--37, 1961

    1961

  25. [33]

    Wang, T., Jiang, Y., and Parkes, D. C. Gemnet: Menu-based, strategy-proof multi-bidder auctions through deep learning. arXiv preprint arXiv:2406.07428, 2024

    2024 arXiv

  26. [34]

    Optimizing multiple performance metrics with deep gsp auctions for e-commerce advertising

    Zhang, Z., Liu, X., Zheng, Z., Zhang, C., Xu, M., Pan, J., Yu, C., Wu, F., Xu, J., and Gai, K. Optimizing multiple performance metrics with deep gsp auctions for e-commerce advertising. In Proceedings of the 14th ACM International Conference on Web Search and Data Mining, pp.\...

    2021

  27. [35]

    Joint auction in the online advertising market

    Zhang, Z., Li, W., Lei, Y., Wang, B., Zhang, Z., Qi, Q., Liu, Q., and Wang, X. Joint auction in the online advertising market. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pp.\ 4362--4373, 2024

    2024

Pith tools

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