Contextual Generative Auction with Permutation-level Externalities for Online Advertising

TL;DR

This paper proposes a novel auction framework, CGA, that models permutation-level externalities in online ad auctions, improving revenue and click-through rates by considering sequential context and externalities.

Contribution

CGA introduces a permutation-level externality model and an end-to-end learning framework, addressing limitations of previous set-level externality approaches.

Findings

CGA outperforms existing methods in offline and online experiments.

CGA achieves nearly optimal revenue with minimal regret.

CGA significantly improves CTR and platform revenue.

Abstract

Online advertising has become a core revenue driver for the internet industry, with ad auctions playing a crucial role in ensuring platform revenue and advertiser incentives. Traditional auction mechanisms, like GSP, rely on the independent CTR assumption and fail to account for the influence of other displayed items, termed externalities. Recent advancements in learning-based auctions have enhanced the encoding of high-dimensional contextual features. However, existing methods are constrained by the "allocation-after-prediction" design paradigm, which models set-level externalities within candidate ads and fails to consider the sequential context of the final allocation, leading to suboptimal results. This paper introduces the Contextual Generative Auction (CGA), a novel framework that incorporates permutation-level externalities in multi-slot ad auctions. Built on the structure of our theoretically derived optimal solution, CGA decouples the optimization of allocation and payment. We construct an autoregressive generative model for allocation and reformulate the incentive compatibility (IC) constraint into minimizing ex-post regret that supports gradient computation, enabling end-to-end learning of the optimal payment rule. Extensive offline and online experiments demonstrate that CGA significantly enhances platform revenue and CTR compared to existing methods, while effectively approximating the optimal auction with nearly maximal revenue and minimal regret.