Online Auction for Ads and Organics

Yudong Hu, Chun Gan, hanminwang, Haoran Wang, Wenlong Chen, Jie He, Changping Peng, Zhangang Lin, Ching Law, Congying Han, Tiande Guo, Fengxiang He

Keywords: online auction, mechanism design, multi-objective optimization

TL;DR: We propose an online blending auction mechanism design approach for sponsored items (ads) with organic items (organics), achieving guaranteed Pareto optimality of platform revenue, advertiser utilities, and user interest.

Abstract: This paper introduces the first online blending auction mechanism design for sponsored items (ads) alongside organic items (organics), ensuring guaranteed Pareto optimality for platform revenue, advertiser utilities, and user interest (measured through clicks). We innovatively define an umbrella term, "traffic item," to encompass both organics and auctionable ad items, where an organic represents a unit of traffic to be auctioned, valued positively by attracting user interest with a fixed zero bid and payment. The online blending traffic distribution problem is thus transformed into an auction problem with unified valuation metric for the traffic item, which is subsequently formulated as an online multi-objective constrained optimization problem. We derive a Pareto equation for this optimization problem, characterizing the optimal auction mechanism set by its solution set. This solution is implemented through a novel two-stage Adaptive Modeled Mechanism Design (AMMD), which (1) trains a hypernetwork to learn a family of parameterized mechanisms, each corresponding to a specific solution of the Pareto equation, and (2) employs feedback-based online control to adaptively adjust the mechanism parameters, ensuring real-time optimality in a dynamic environment. Extensive experiments demonstrate that AMMD outperforms existing methods in both click-through rates and revenue across multiple auction scenarios, particularly highlighting its adaptability to online environments. The code has been submitted and will be released publicly.

Primary Area: other topics in machine learning (i.e., none of the above)

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 7070