Profile
Public facts about Hong Ting Tsang (曾康庭, Dennis), a PhD student in computer science at HKUST working on enterprise data agents, agent memory, knowledge graphs, and large language models.
Hong Ting Tsang (曾康庭, Dennis) is a first-year PhD student in computer science at the Hong Kong University of Science and Technology, advised by Prof. Yangqiu Song. He studies enterprise data agents, agent memory, knowledge graphs, and large language models.
This page restates the public facts from this site in one place. The personal homepage is gzone0111.github.io. Last updated 5 November 2026.
Identity
Hong Ting Tsang (曾康庭, Dennis) is a PhD student in the Department of Computer Science and Engineering at the Hong Kong University of Science and Technology (HKUST). Prof. Yangqiu Song is his advisor. Before the PhD, he completed an MPhil and a Bachelor of Engineering in Computer Science at HKUST. He is based in Hong Kong. Email: httsangaj@cse.ust.hk.
Research
His research covers four areas:
- Enterprise data agents: agent memory, LLM agents, and their applications.
- Knowledge bases: knowledge graph construction and reasoning.
- Large language models: large-scale language-model applications over big data.
- Data mining: methods for structured and unstructured data.
Publications
Three papers are featured on the homepage: MOSAIC, AutoGraph-R1, and AutoSchemaKG. Venues below are the ones recorded on the publications page. A paper listed as arXiv is a preprint, not an acceptance.
| Paper | Venue | Year | Link |
|---|---|---|---|
| MOSAIC: An Omni-Modal Benchmark for Long-Term Memory and Cross-Modal Reasoning in Conversational Agents | arXiv | 2026 | 2610.29249 |
| AutoGraph-R1: End-to-End Reinforcement Learning for Knowledge Graph Construction | ACL (Main) | 2026 | 2510.15339 |
| AutoSchemaKG: Autonomous Knowledge Graph Construction through Dynamic Schema Induction from Web-Scale Corpora | ACL (Main, Oral) | 2026 | 2505.23628 |
| AtlasKV: Augmenting LLMs with Billion-Scale Knowledge Graphs in 20GB VRAM | ICLR | 2026 | OpenReview |
| The Cognitive Bandwidth Bottleneck: Shifting Long-Horizon Agents from Planning with Actions to Planning with Schemas | EMNLP | 2026 | Anthology |
| ContextLens: Modeling Imperfect Privacy and Safety Context for Legal Compliance | ACL (Main) | 2026 | 2604.12308 |
| DeepRefine: Agentic Knowledge Refinement via Reinforcement Learning | arXiv | 2026 | 2605.10488 |
| Can Language Models Imagine Without Seeing? Ekphrasis: Measuring Visual Creative Ideation in Text-Only LLMs | arXiv | 2026 | 2608.06967 |
| Towards In-Parameter Memory Augmentation for Large Language Models | arXiv | 2026 | 2610.08630 |
| KGPFN: Unlocking the Potential of Knowledge Graph Foundation Model via In-Context Learning | arXiv | 2026 | 2605.14907 |
| NewtonBench: Benchmarking Generalizable Scientific Law Discovery in LLM Agents | ICLR | 2026 | 2510.07172 |
| NGDB-Zoo: Towards Efficient and Scalable Neural Graph Databases Training | NeurIPS | 2026 | 2602.21597 |
| PatchWorld: Gradient-Free Optimization of Executable World Models for Agent Environments | arXiv | 2026 | 2605.30880 |
| Transformers for Complex Query Answering over Knowledge Hypergraphs | arXiv | 2025 | 2504.16537 |
| From Automation to Autonomy: A Survey on Large Language Models in Scientific Discovery | EMNLP | 2025 | 2505.13259 |
| TERAG: Token-Efficient Graph-Based Retrieval-Augmented Generation | ICMLC | 2025 | 2509.18667 |
| Top Ten Challenges Towards Agentic Neural Graph Databases | IEEE Data Engineering Bulletin | 2025 | 2501.14224 |
| NGDBench: Towards Neural Graph Data Management | Findings of EMNLP | 2026 | Anthology |
The bibliographic source is publications. Google Scholar: Hong Ting Tsang. ORCID: 0009-0004-3045-2566.
Teaching
He was a teaching assistant at HKUST CS for two courses:
- COMP 4222, Machine Learning with Structured Data, 2025–26 Fall. Course note.
- COMP 4332, Big Data Mining and Management, 2024–25 Spring. Course note.
Notes
Reading notes on this site:
- The Society of Mind, 10 April 2026.
- GoF Design Patterns, 10 April 2026.
- Hamming — You and Your Research, 7 March 2026.
Profiles
- Email: httsangaj@cse.ust.hk
- Advisor: Prof. Yangqiu Song
- Google Scholar: Bpk3L20AAAAJ
- ORCID: 0009-0004-3045-2566
- GitHub: gzone0111
- LinkedIn: hong-ting-tsang-dennis
- X: gzone0111
Frequently asked questions
Who is Hong Ting Tsang?
Hong Ting Tsang (曾康庭, Dennis) is a first-year PhD student in computer science at the Hong Kong University of Science and Technology. Prof. Yangqiu Song is his advisor. He previously completed an MPhil and a Bachelor of Engineering in Computer Science at HKUST.
What does Hong Ting Tsang research?
He studies enterprise data agents, agent memory, large language models, knowledge bases, and data mining. The current focus is enterprise data agents, agent memory, knowledge graph construction and reasoning, large-scale language-model applications over big data, and data mining for structured and unstructured data.
Which papers should I read first?
Start with MOSAIC, an omni-modal benchmark for long-term memory in conversational agents; AutoGraph-R1, published at ACL 2026 (Main); and AutoSchemaKG, an oral presentation at ACL 2026 (Main). NGDB-Zoo was published at NeurIPS 2026. The full list is in the publications table on this page.
Where has Hong Ting Tsang taught?
He was a teaching assistant for COMP 4222, Machine Learning with Structured Data, at HKUST CS in 2025–26 Fall, and for COMP 4332, Big Data Mining and Management, at HKUST CS in 2024–25 Spring.