Our lab would like to thank generous support from The Hong Kong Polytechnic University, The Transport Department, Hong Kong Productivity Council, and University Grants Committee.
Ongoing Research Projects
Mixture of Expert Graphs for Precise Retrieval-Augmented Generation, Jan 2027 - Dec 2029, CNY300,000, NSFC, PI
Tiered-Competency Personalized Learning and Assessment Platform, Innovation Education Development Scheme, Oct 2026-Sep 2028, HK$1,200,000, PI
Self-Paced Learning Platform Powered by Efficient Graph Retrieval-Augmented Generation, Teaching Development Grant, Jan 2026–Dec 2027, HK$1,084,618, PI
Trustworthy Response Generation in Large Language Models via Multimodal Knowledge Fusion, PolyU Distinguished Postdoctoral Fellowship Scheme, Dec 2025 - 30 Nov 2027, HK$841,040, PI.
Operation and Maintenance Copilot based on Graph Retrieval-Augmented Generation, Innovation and Technology Fund, Nov 2025–Nov 2027, HK$1,050,900, PI
Cross-Correlated Graph Neural Networks with Applications in Large-Scale Recommender Systems, Innovation and Technology Fund, Mar 2025–Mar 2027, HK$1,516,999.4, PI
Intelligent Virtual Assistant for Operation and Maintenance based on Large Language Models, Donation from the Industry, Jul 2024–Dec 2026, HK$660,000, PI
DataChat: Chatting with Database based on Knowledge-Graph-Enhanced Large Language Models, Departmental Entrepreneurship Fund, Nov 2023–Nov 2026, HK$500,000, PI
Intelligent and Trusted FinTech Research Thrust, Sep 2026-Aug 2029, Impactful Research Hub Fund, HK$6,204,600, Co-I
Designing Algorithmic Aesthetics: Evaluating AI Performance in Applying Diverse Interior Styles to Actual 3D Spatial Reconstructions, ADO Limited (Donation), Jul 2026 - 31 Dec 2029, HK$260,870, Co-I
Collaborative Generative AI (Co-GenAI), Theme-based Research Scheme, Jan 2026–Dec 2030, HK$62.606 million, Co-I
Development of a Hierarchical Thermal Dynamic Modelling Method Based on Physically Consistent GNN for High-Rise Buildings, General Research Fund, Jan 2026–Dec 2028, HK$1,169,801, Co-I
Develop a Vision-Language Model-based Smart Driving Assistant for Enhancing Safety and Convenience of Motorists, Smart Traffic Fund, Mar 2025–Feb 2027, HK$5,012,789.42, Co-I
Course Content Knowledge Graph Construction and Reasoning Powered by LLMs, Departmental Research, Sep 2024–Aug 2027, HK$1,614,620, Co-I
Image-based Unsupervised 3D Scene Understanding and Editing, NSFC General Program, Jan 2023–Dec 2026, CNY592,000, Co-I
Completed Research Projects
Error-Aware and Contrastive Knowledge Graph Learning, Early Career Scheme (RGC), Jan 2023–Dec 2025, HK$908,800, PI
Human-Centric Attributed Network Learning, Start-up Fund for New Recruits, Aug 2020–Aug 2023, HK$1,000,000, PI
Integrating Contrastive Error Detection and Accelerated Completion for Large-Scale Knowledge Graph Refinement, Departmental General Research Fund, Jul 2021–Jul 2023, HK$300,000, PI
Design and formulation of behavioural user profile of teenage sports player with AI model and analysis methods from basketball training, SD/COMP Joint Research Scheme, Sep 2022–Aug 2024, HK$500,000, co-PI
Parking Garage Vacancy Prediction Services in Hong Kong: AI-enabled Solutions for Enhanced Reliability and Extensibility, Smart Traffic Fund, Mar 2025–Aug 2026, HK$2,897,745.06, Co-I
Cognition-aware 3D Content Design for User-centered Metaverses, SD/COMP Joint Research Scheme, Sep 2022–May 2026, HK$500,000, Co-I
Development of an Augmented Reality-Assisted Head-up Display (AR-HUD) mechanism for recommending driving strategy, Smart Traffic Fund, Jan 2023–Dec 2024, HK$1,315,127, Co-I
Development and Deployment of an AI-enabled Parking Vacancy Prediction Framework using Multi-source Data, Smart Traffic Fund, May 2022–Jul 2023, HK$985,034, Co-I
Research Project Statistics
Number of Projects
| Role | Ongoing | Completed | Total |
| PI and co-PI | 8 | 4 | 12 |
| Co-I | 7 | 4 | 11 |
| All | 15 | 8 | 23 |
Disclosed Funding
| Funding Category | Ongoing | Completed | Total |
| All projects as PI and co-PI (HKD) | 6,853,557.40 | 2,708,800.00 | 9,562,357.40 |
| All projects as Co-I (HKD) | 76,868,680.42 | 5,697,906.06 | 82,566,586.48 |
| All projects as PI and co-PI (CNY) | 300,000.00 | – | 300,000.00 |
| All projects as Co-I (CNY) | 592,000.00 | – | 592,000.00 |
Research Focus
The effective modeling of graph data is crucial for advancing AI, as it underpins systems from social networks to recommendation engines. My research portfolio addresses this need through the development of innovative graph learning algorithms and tools, which have demonstrated state-of-the-art performance in practical tasks such as classification, anomaly detection, and recommendations. A key focus has been applying graph learning to enhance the reasoning capabilities of large language models (LLMs) for applications such as question answering and SQL generation.
1. Retrieval-Augmented Generation
2. LLM-based Text-to-SQL
3. Graph Learning
Each Graph is a New Language: Graph Learning with LLMs, Huachi Zhou, Jiahe Du, Chuang Zhou, Chang Yang, Yilin Xiao, Yuxuan Xie, Xiao Huang, ACL (Findings), 2025. (Code)
Dirichlet Energy Constrained Learning for Deep Graph Neural Networks, Kaixiong Zhou, Xiao Huang, Daochen Zha, Rui Chen, Li Li, Soo-Hyun Choi, Xia Hu, NeurIPS, 2021.
Towards Deeper Graph Neural Networks with Differentiable Group Normalization, Kaixiong Zhou, Xiao Huang, Yuening Li, Daochen Zha, Rui Chen, Xia Hu, NeurIPS, 2020.
4. Collaborative Filtering
