How Deep Learning Is Revolutionizing E‑Commerce Search and Chatbots

Deep Learning at QCon 2017

On April 16‑18, 2017, the QCon Beijing conference featured a keynote by Jim Cheng, director of Suning’s U.S. Silicon Valley Research Institute, titled “Deep Learning in E‑Commerce.”

Jim explained that while deep learning has already excelled in image and speech recognition, its application to natural language processing is just beginning, especially for e‑commerce scenarios such as product search and conversational assistants.

Search relevance : He described a “vector search engine” that converts product documents into numeric vectors, enabling the system to match user intent rather than relying on simple keyword matching. Existing word‑embedding methods are mature, but sentence‑ and document‑level vectorization remains an open challenge.

Conversational commerce : By integrating deep‑learning models with traditional rule‑based approaches, the team built a chat‑bot platform that can interpret user utterances and images using only a modest amount of manually labeled data, turning chat interfaces into a new entry point for online shopping.

Jim, a Stanford PhD and AI application specialist, leads a research team founded in 2013 that brings together experts in big data, artificial intelligence, and deep learning. Their work powers Suning’s smart‑retail ecosystem, including AI‑driven search, VR/AR experiences, and intelligent home solutions.

The institute continues to explore frontier technologies to create unified user and enterprise profiles, driving personalization and establishing Suning as a technology‑driven commercial leader.