Donglin Zhou
Donglin Zhou
I’m a Recommendation Algorithm Engineer at Vipshop, where I work on recommender systems. I received my M.S. from the College of Computer Science and Software Engineering, Shenzhen University, advised by Prof. Weike Pan, and my B.E. in Computer Science and Technology from Shantou University, where I was advised by Prof. Lin Zheng. My research lies at the intersection of recommender systems and large language models, with a focus on LLM-based recommendation, multimodal item representation, generative recommendation, and LLM evaluation. Previously, I worked on Hunyuan LLM evaluation at Tencent TEG and local-life recommendation at Kuaishou.
EmailGitHubGoogle ScholarLinkedInCVBlog

News

Research Highlights

I write ongoing paper readings and research notes at RecSys & LLM Frontier Notes.

Education

2023.09 - 2026.06

M.S., College of Computer Science and Software Engineering, Shenzhen University

Computer Science and Technology; advised by Prof. Weike Pan. My SIGIR 2026, KDD 2025 and FCS recommendation research was conducted during my master's studies.

2019.09 - 2023.06

B.E. in Computer Science and Technology, Shantou University

Advised by Prof. Lin Zheng.

Work Experience

2026.07 - Present

Recommendation Algorithm Engineer, Vipshop

Recommender Systems

Working on recommendation algorithms for e-commerce.

2025.04 - 2025.11

Recommendation Algorithm Engineer Intern, Kuaishou Technology

Commercialization Algorithm · Local-life Short-video Recommendation
  • Designed LLM-aligned semantic item IDs using co-occurrence relations and RQ-VAE, and integrated them into an MMoE ranking model.
  • Developed geographic-density-based sampling and LLM-aligned semantic embeddings for POI tokenization and behavior sequence modeling; this research later appeared at AAAI 2026.
2023.06 - 2023.08

Data Engineering / Algorithm Intern, Tencent TEG

Data Platform Department · LLM Evaluation Team
  • Evaluated Hunyuan LLM across NLP fundamentals, generation, multi-turn dialogue, reasoning, domain knowledge and safety.
  • Conducted the TencentLLMEval work in August 2023, including GPT-4 based automatic evaluation prompts and a three-level task taxonomy covering 7 domains, 200+ categories and 800+ tasks; the paper was published in ACM TIST in August 2026.

Selected Publications

2026
LLM-based Semantic and ID Representations for Sequential Recommendation
Donglin Zhou, Weike Pan, Zhong Ming
SIGIR 2026 · CCF-AProceedings of the 49th International ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 4403-4408

Studies how LLM-derived semantic representations and collaborative ID representations can be jointly modeled for sequential recommendation.

DOI · Scholar · CodeCitations: 0 (Google Scholar, 2026-09-19)
2026
LLM-Aligned Geographic Item Tokenization for Local-Life Recommendation
Hao Jiang, Guoquan Wang, Donglin Zhou, Sheng Yu, Yang Zeng, Wencong Zeng, Kun Gai, Guorui Zhou
AAAI 2026 · CCF-AProceedings of the AAAI Conference on Artificial Intelligence, 40(17), pp. 14928-14936

Introduces geographic item tokens for local-life recommendation, connecting semantic embeddings with spatial and collaborative signals.

DOI · Scholar · CodeCitations: 7 (Google Scholar, 2026-09-19)
2026
TencentLLMEval: A Hierarchical Evaluation of Real-World Capabilities for Human-Aligned LLMs
Shuyi Xie, Wenlin Yao, Yong Dai, Shaobo Wang, Zishan Xu, Fan Lin, Donglin Zhou, et al.
ACM TIST · JCR Q1ACM Transactions on Intelligent Systems and Technology, 17(4), pp. 1-22

Builds a hierarchical real-world benchmark and studies human-aligned, LLM-as-a-judge evaluation.

DOI · ScholarCitations: 10 (Google Scholar, 2026-09-19)
2025
Contrastive Text-enhanced Transformer for Cross-Domain Sequential Recommendation
Donglin Zhou, Xinbei Cai, Weike Pan
KDD 2025 · CCF-AProceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pp. 4110-4119

Transfers text-enhanced preferences across domains by combining textual semantics, collaborative signals and contrastive learning.

DOI · Scholar · CodeCitations: 8 (Google Scholar, 2026-09-19)
2025
Self-Supervised Representation Learning with ID-Content Modality Alignment for Sequential Recommendation
Donglin Zhou, Weike Pan, Zhong Ming
Frontiers of Computer ScienceFrontiers of Computer Science, 20(12)

Aligns ID, text and image signals through self-supervised learning for multimodal sequential recommendation.

DOI · Scholar · CodeCitations: 1 (Google Scholar, 2026-09-19)
2023
Attenuated Sentiment-aware Sequential Recommendation
Donglin Zhou, Zhihong Zhang, Yangxin Zheng, Zhenting Zou, Lin Zheng
IJDSA 2023International Journal of Data Science and Analytics, 16(2), pp. 271-283

Models time-attenuated sentiment signals extracted from reviews for sequential recommendation.

DOI · Scholar · CodeCitations: 8 (Google Scholar, 2026-09-19)

Open-source Projects