Large Models and Recommendation Systems: Challenges, Opportunities, and Industry Insights (CNCC 2023 Technical Forum)

In the era of information explosion, recommendation systems play a key role in alleviating information overload. From early systems that simply recommended items based on user searches, they have evolved into highly personalized, "mind‑reading" engines that understand users more deeply.

Large models bring unprecedented opportunities to recommendation systems. Their strong generalization and knowledge‑understanding abilities can deliver more accurate recommendations and better user experiences, while also introducing challenges such as high computational resource demands, model interpretability, and ethical considerations.

The CNCC 2023 Technical Forum, held on October 27, 2023 from 13:30 to 17:30, gathers leading experts to discuss “Large Models and Recommendation Systems.” The event is organized by Xiaohongshu and features speakers including Xiaohongshu Vice President Feng Di, USTC Professor He Xiangnan, Tsinghua University Professor Zhang Min, Renmin University Professor Zhao Xin, and Huawei Noah’s Ark Lab Director Tang Ruiming.

Schedule and abstracts:

13:30‑14:00 – Challenges and Opportunities of Recommendation Systems in the Era of Large Models (Zhang Min, Tsinghua University). The talk reviews recent advances, research directions, and valuable topics that cannot be ignored in the large‑model era.

14:00‑14:30 – Innovative Exploration of Xiaohongshu Recommendation System (Feng Di, Xiaohongshu). It covers cold‑start problems, multimodal content understanding, multi‑objective ranking, E&E methods, and agent‑based search‑recommendation fusion, highlighting industrial practice and future directions.

14:30‑15:00 – Frontiers and Outlook of Large Model Recommendation (He Xiangnan, USTC). The presentation introduces LLM‑based recommendation techniques such as in‑context learning, instruction‑tuning, and generative recommendation, and discusses associated opportunities and challenges.

15:00‑15:30 – Application of Large Language Models in Recommendation Systems (Zhao Xin, Renmin University). It outlines the main capabilities of LLMs, their impact on information retrieval and recommendation, and explores future research paradigms.

15:30‑16:00 – How Recommendation Systems Complement Large Language Models: An Application Perspective (Tang Ruiming, Huawei). The talk examines semantic signals and external knowledge from LLMs, the lack of collaborative signals, latency issues, and proposes two frameworks for integrating LLMs with recommendation pipelines.

16:00‑17:30 – Panel discussion “Large Models Empower Recommendation Systems: Opportunities and Challenges” with all speakers, delving deeper into technical advantages, limitations, and future directions.

The live stream will be broadcast on WeChat Channels (Xiaohongshu Tech REDtech), Bilibili, and Douyin. Participants can register via QR code and receive gifts such as electric toothbrushes and desk lamps. The forum is part of the China Computer Conference (CNCC 2023) held in Shenyang, with an offline exhibition booth (B79) for further interaction.