This article summarizes a AAAI 2021 paper that introduces a deep meta‑learning framework with an amortization network to generate spatio‑temporal representations, enabling accurate retail sales predictions across regions and time periods, especially during high‑volume shopping festivals.
The presentation details the design and implementation of JD’s Urban Spatio‑Temporal Data Engine (JUST), explaining its architecture, novel storage and indexing techniques, performance optimizations, experimental results, and real‑world applications such as pandemic contact tracing and hazardous‑material monitoring, while highlighting its academic impact.
The presentation details the design, implementation, and real‑world applications of the JD Urban Spatio‑Temporal Data Engine (JUST), a distributed, scalable database that handles massive, complex spatio‑temporal data with novel storage, indexing, and query techniques, demonstrating high performance and ease of use across smart‑city scenarios.
The paper proposes a deep meta‑learning framework that generates spatio‑temporal representations for retail sales forecasting, especially during large shopping festivals, by combining amortization networks, shared statistical structures, and alternating spatial‑temporal training to achieve robust and accurate predictions despite scarce historical data.
The article presents a novel spatio‑temporal trajectory risk metric and an efficient query framework, including a space‑first index (SFT) and distributed storage, to identify high‑risk contacts of COVID‑19 patients from massive movement data, with real‑world deployment results in several Chinese cities.
