Yijie Zhong
Ph.D. Student at Tongji University
dun.haski@gmail.com zhongyj@tongji.edu.cn
I am a Ph.D. student at the Knowledge Computing Lab, Tongji University, supervised by Prof. Haofen Wang. I received my M.S. degree from Nanjing University in 2022, under the supervision of Prof. Zhengxing Sun. My research interests encompass information extraction, memory and retrieval mechanisms in Retrieval-Augmented Generation (RAG) and LLM Agents, with a particular focus on personal knowledge graph construction, personal information management, and personalized applications on intelligent devices. To date, I have published 16 papers in top-tier conferences and journals, 10 of which are first-authored. An additional 5 papers are currently under review.
News
2026.04
Paper accepted to SIGIR 2026 (CCF-A)
2026.02
Paper accepted to CVPR 2026 (CCF-A)
2026.01
Paper accepted to WWW 2026 (CCF-A)
2026.01
Paper accepted to KBS (CAS Q1)
2025.11
Paper accepted to TOIS (CAS Q1)
2025.08
Started internship at vivo BlueImage Lab
Education
2022 - Present
Tongji University · Ph.D.
College of Design and Innovation, AI-Design (Engineering), Advisor: Prof. Haofen Wang
QS 2026 Subject Ranking: 16th Globally, 2nd in Asia
QS 2026 Subject Ranking: 16th Globally, 2nd in Asia
2019 - 2022
Nanjing University · M.S.
Department of Computer Science and Technology, Computer Science and Technology, Advisor: Prof. Zhengxing Sun
National Key Laboratory
National Key Laboratory
2015 - 2019
Nanjing University of Aeronautics and Astronautics · B.S.
School of Computer Science and Technology, Information Security
Research Interests
Now:
- RAG
- LLM & MLLM Agent
- Memory Mechanism
- Information Extraction
- Personal Knowledge Management
Previous:
- Salient & Camouflaged Object Detection
- Image Matting
- 3D Reconstruction
Publications
For a full list, please see publications page
CAPTAIN: Cognitive Visual Profiling and Intent Navigation for Fine-grained Image Retrieval and Reasoning
IEEE Transactions on Multimedia (TMM), Under Review
Hierarchical Textual Knowledge for Enhanced Image Clustering
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026
Scene-Aware Memory Discrimination: Deciding Which Personal Knowledge Stays
Knowledge-Based Systems (KBS), 2026
HingeMem: Boundary Guided Long-Term Memory with Query Adaptive Retrieval for Scalable Dialogues
The Web Conference (WWW), 2026
ODDA: An OODA-Driven Diverse Data Augmentation Framework for Low-Resource Relation Extraction
Annual Meeting of the Association for Computational Linguistics (ACL), 2025
Meta-PKE: Memory-Enhanced Task-Adaptive Personal Knowledge Extraction in Daily Life
Information Processing & Management (IPM), 2025
Detecting Camouflaged Object in Frequency Domain
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022
Experience
2025.08 - Present · Shanghai
vivo Imaging Algorithm Research Lab, BlueImage Lab
Research Intern
- Explored RAG-based question answering and image captioning for photo album scenarios; designed a systematic image-to-text algorithm and built a textual memory bank to improve efficiency while comprehensively covering visual information
- Investigated image quality assessment methods and comparison scheme design
2023.09 - 2024.09 · Shenzhen, Guangdong
Huawei 2012 Laboratories, Central Research Institute, Planning Department
Research Intern
- Participated in the construction of the "Personal Data Foundation", exploring the transformation of multi-source heterogeneous user data from terminal devices into usable personal knowledge bases; identified end-to-end use cases transitioning from passive user input to proactive device triggering, and from passive application initiation to context-aware activation
- Designed personal schema and knowledge extraction algorithms for text data (SMS, email, memos, etc.) and built automated data construction pipelines; developed person intimacy mining and event detection algorithms for photo album images
- Explored memory write, storage, and retrieval mechanisms for massive input and interaction data, reducing noise, avoiding redundant computation, improving storage efficiency and retrieval accuracy, while supporting multi-type queries and diverse scenario requirements
- Collaborated closely with Celia, Theoretical Department, Advanced Computing and Storage Laboratory, and Terminal BG Pre-research Department to drive technology deployment; algorithms ultimately became trade secrets and entered Harmony 5.0 concept project and Harmony 5.1 technology pipeline
2018.06 - 2018.09 · Nanjing, Jiangsu
Huawei Nanjing Research Institute, Software Product Line
Intern
- Streamlined end-to-end business logic and data flow for the communications product channel sales platform across front, middle, and back offices
- Independently designed and developed new front-end components for emerging business requirements
Talks
2025.12
OpenKG Ph.D. Student Forum · Nanjing
Talk: Personal Knowledge Extraction and Memory Management: Selection, Construction, and Application
2000+ attendees (in-person & remote)
2000+ attendees (in-person & remote)
2025.09
SJTU 3A Alumni Academic Exchange Session 23 · Ph.D. Forum · Shanghai
Talk: Efficient Construction of Personal Memory Systems: From Memory Discrimination to Retrieval-Augmented Memory
Academic Service
Reviewed nearly 100 papers in total, and assisted in reviewing dozens of master's and doctoral theses as well as project application materials.