A Joint Auction Framework with Externalities and Adaptation

TL;DR

This paper tackles the challenge of optimizing revenue and user experience in online advertising where joint advertising (store-brand bundles) coexists with traditional ads. It introduces JEANet, an Automated Mechanism Design-based neural architecture composed of Adaptive Extraction Module, Externality Transformer Module, and Deep Mechanism Module to generate globally optimal hybrid ad lists under externalities and bid heterogeneity. JEANet enforces DSIC and IR while allowing integration of joint and traditional advertising, and it explicitly models cross-item externalities and distributional differences among advertisers. Extensive experiments on synthetic and industrial data demonstrate that JEANet outperforms state-of-the-art baselines in revenue and user experience, with feasible training and rapid online decision-making, highlighting its practical applicability for real-world platforms.

Abstract

Recently, joint advertising has gained significant attention as an effective approach to enhancing the efficiency and revenue of advertising slot allocation. Unlike traditional advertising, which allocates advertising slots exclusively to a single advertiser, joint advertising displays advertisements from brands and stores that have established a joint selling relationship within the same advertising slot. However, existing approaches often struggle to accommodate both joint and traditional advertising frameworks, thereby limiting the revenue potential and generalizability of joint advertising. Furthermore, these methods are constrained by two critical limitations: they generally neglect the influence of global externalities, and they fail to address the bidding variability stemming from multi-party advertiser participation. Collectively, these limitations present substantial challenges to the design of joint auction mechanisms. To address these challenges, we propose a Joint Auction Framework incorporating Externalities and Adaptation, and leverage the automated mechanism design (AMD) method through our proposed JEANet to compute joint auction mechanisms that satisfy the conditions of individual rationality (IR) and approximate dominant strategy incentive compatibility (DSIC). As the first AMD method to integrate global externalities into joint auctions, JEANet dynamically adapts to the bidding characteristics of multi-party advertiser and enables unified auctions that integrate both joint and traditional advertising. Extensive experimental results demonstrate that JEANet outperforms state-of-the-art baselines in multi-slot joint auctions.

Figures (3)

• Figure 1: Histogram of bid differences among store, brand, and joint advertisers. For comparison, this figure illustrates the normalized bid distribution of different advertisers.
• Figure 2: JEANet is the first AMD method that explicitly accounts for externalities in joint auctions. Numerous studies have employed the AMD method to analyze externalities in traditional and joint advertising. However, no research has explored the impact of externalities in joint advertising.
• Figure 3: The architecture of JEANet. JEANet consists of three modules: Adaptive Extraction Module (AEM), Externality Transformer Module (ETM) and Deep Mechanism Module (DMM).

Theorems & Definitions (2)

• definition 1: Dominant Strategy Incentive Compatibility
• definition 2: Individual Rationality