AIDA Advertising Intelligent Decision and Allocation Framework: Evolution of Smart Auction Mechanisms

The presentation, led by senior algorithm expert Zhang Zhilin from Alibaba, outlines the rapid growth of the online advertising market and the central role of mechanism design in allocating valuable traffic to advertisers while ensuring platform revenue and stability.

It first describes the Alibaba Mama display advertising business, its core products (Super Recommendation, Super Diamond), the full‑chain consumer targeting across Alibaba’s internal and external media, and the diverse marketing goals of advertisers that drive the need for intelligent bidding and multi‑objective optimization.

The current platform uses a platform‑centric OCPC approach, which couples multiple stakeholder objectives and suffers from inefficiencies; this motivates the development of a new, future‑oriented framework called AIDA.

AIDA consists of a two‑layer collaborative algorithmic architecture: a lower‑layer bidding agent that monetizes traffic for each advertiser’s multiple goals, and an upper‑layer auction mechanism that ensures economic properties such as incentive compatibility and Nash equilibrium.

Building on Myerson’s theorem and the classic GSP model, the team designed Deep GSP, a deep‑learning‑enhanced auction that guarantees monotonic allocation, minimal payment, and multi‑objective optimization, achieving superior Pareto performance and significant ROI gains in production.

To overcome limitations of Deep GSP, the Neural Auction was introduced, employing a differentiable sorting engine, set encoder, and context‑aware rank‑score function to enable end‑to‑end learning of auction mechanisms directly from real‑world feedback.

The AIDA platformization includes a graph‑based online service engine (AIDA agent) built on TensorFlow for high‑concurrency serving, and an offline strategy solution integrated into Alibaba’s Star Cloud platform, accelerating the iteration of mechanism strategies.

Future plans focus on deeper technical research, new scenarios and technologies, ecosystem optimization, fundamental mechanism theory, and open‑source platformization.

The article concludes with a recruitment call for machine‑learning talent (interns and full‑time) to join the Alibaba Mama advertising team, highlighting the large‑scale data, business impact, and cutting‑edge research opportunities.