The paper presents a tiered NVMe SSD solution for ad‑ranking retrieval, using zone‑aware hardware, Direct I/O with libaio/IoUring, and an adaptive pipeline scheduler to cut long‑tail read latency and boost throughput, enabling larger‑than‑memory signals and improving ranking accuracy and revenue.
SMAD is a scalable graph‑based ad retrieval framework for e‑commerce search that builds a heterogeneous Query‑Item‑Ad graph, learns multi‑view embeddings with a parallel deep neural network and attention, employs category‑aware sampling for efficient distributed training, and delivers significant gains in offline relevance and online CTR, RPM, and PVR.
Alibaba’s latest AI-driven ad retrieval framework, unveiled at WWW 2018, replaces keyword‑based search with a user‑behavior heterogeneous graph and machine‑learning models, delivering personalized, high‑efficiency ad matching that boosts ROI for advertisers, improves user experience, and enhances platform revenue.
