📝 Publications

A Survey of Task-Oriented Knowledge Graph Reasoning: Status, Applications, and Prospects
Guanglin Niu, Bo Li, Yangguang Lin
Arxiv, 2025

• This survey aims to highlight key research trends and outline promising future directions of Knowledge Graph Reasoning (KGR) from the perspective of primary reasoning tasks, downstream applications, and challenging tasks.
• We explore advanced techniques, such as large language models, and their impact on KGR.
• Our work is promoted by some media, such as 人工智能遇上知识图谱.

📃Paper 🔧Project

Diffusion-based Hierarchical Negative Sampling for Multimodal Knowledge Graph Completion
Guanglin Niu, Xiaowei Zhang
The 30th International Conference on Database Systems for Advanced Applications (DASFAA), 2025

• DHNS is the first to leverage the diffusion model’s capabilities within the context of multi-modal knowledge graph for negative sampling.

📃Paper 💾Code

Logic and Commonsense-Guided Temporal Knowledge Graph Completion
Guanglin Niu, Bo Li
Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2023

• This work LCGE is the first to introduce temporal rules into temporal knowledge graph completion models.
• LCGE models each event from the perspectives of both the time-sensitive representation and the commonsense.
• Our work is promoted by some forums, such as AI Time青年科学家论坛.

📃Paper 💾Code

CAKE: A Scalable Commonsense-Aware Framework For Multi-View Knowledge Graph Completion
Guanglin Niu, Bo Li, Yongfei Zhang, Shiliang Pu
Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) (ACL), 2022

• This work CAKE is the first to propose a scalable knowledge graph completion framework to predict entities in a joint commonsense and fact-driven fashion.
• CAKE generates commonsense automatically for negative sampling and multi-view link prediction.
• Our work is promoted by several media and forums, such as AI Time 视频 | AI Time 解读、专知、智源社区、开放知识图谱、AMiner.

📃Paper 💾Code

Relational Learning with Gated and Attentive Neighbor Aggregator for Few-Shot Knowledge Graph Completion
Guanglin Niu, Yang Li, Chengguang Tang, Ruiying Geng, Jian Dai, Qiao Liu, Hao Wang, Jian Sun, Fei Huang, Luo Si
Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR), 2021

• This approach GANA is the first to propose a gated and attentive neighbor aggregator to capture the most valuable contextual semantics of a relation.
• GANA is one of the most representative models and always selected as the baseline on few-show knowledge graph completion tasks.
• This work was conducted in collaboration with Qwen team. Our work is promoted by some media and forums, such as 专知.

📃Paper 💾Code

Rule-Guided Compositional Representation Learning on Knowledge Graphs
Guanglin Niu, Yongfei Zhang, Bo Li, Peng Cui, Si Liu, Jingyang Li and Xiaowei Zhang
Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2020

• This work RPJE is the first attempt to integrate logic rules with paths for KG embedding, balancing the explainability and the generalization.
• Our work is promoted by several media and forums, such as 开放知识图谱、雷锋网、SAAI、MLNLP. Particularly, RPJE was recognized as one of the representative studies of neuro-symbolic KG reasoning at (CCKS 2021).

📃Paper 💾Code

AutoETER: Automated Entity Type Representation with Relation-Aware Attention for Knowledge Graph Embedding
Guanglin Niu, Yongfei Zhang, Bo Li, Shiliang Pu and Jingyang Li
Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: Findings (EMNLP Findings), 2020

• This work is the first to automatically learn the embeddings of entity types to enrich the general features of entities without explicit type information.
• Our work is promoted by some media, such as AI Time、SFFAI、专知.

📃Paper 💾Code

Perform like an Engine: A Closed-Loop Neural-Symbolic Learning Framework for Knowledge Graph Inference
Guanglin Niu, Bo Li, Yongfei Zhang, Shiliang Pu
Proceedings of the 29th International Conference on Computational Linguistics (COLING), 2022

• EngineKG performs like a four-stroke engine in a closed-loop neural-symbolic learning framework with embedding-based rule learning and rule-enhanced knowledge graph embedding.
• Our work is promoted by some media and forums, such as MLNLP Talk.

📃Paper 💾Code

From Poses to Identity: Training-Free Person Re-Identification via Feature Centralization
Chao Yuan, Guiwei Zhang, Changxiao Ma, Tianyi Zhang, Guanglin Niu(Corresponding Author)
Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025

• Pose2ID is a Training-Free Feature Centralization framework that can be directly applied to different ReID tasks and models, even an ImageNet pre-trained model without ReID training.
• The developed Identity-Guided Pedestrian Generation paradigm leverages identity features to generate high-quality images of the same identity in different poses to achieve feature centralization.

📃Paper 💾Code

Try Harder: Hard Sample Generation and Learning for Clothes-Changing Person Re-ID
Hankun Liu, Yujian Zhao, Guanglin Niu(Corresponding Author)
Proceedings of the 33rd ACM International Conference on Multimedia (ACM MM), 2025

• HSGL is a novel multimodal-guided Hard Sample Generation and Learning framework, which is the first effort to unify textual and visual modalities to explicitly define, generate, and optimize hard samples within a unified paradigm.
• We propose a hardness-aware optimization strategy that adjusts feature distances based on textual semantic labels, encouraging the separation of hard positives and drawing hard negatives closer in the embedding space to enhance the model’s discriminative capability and robustness to hard samples.

📃Paper 💾Code

CAMEL: CAusal Motion Enhancement tailored for Lifting Text-driven Video Editing
Guiwei Zhang, Tianyu Zhang, Guanglin Niu(Corresponding Author), Zichang Tan, Yalong Bai, Qing Yang
Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024

• CAMEL develops a causal motion-enhanced attention mechanism to enhance the motion coherence of latent representations while preserving content generalization to creative textual scenarios.

📃Paper 💾Code

Identity-aware Feature Decoupling Learning for Clothing-change Person Re-identification
Haoxuan Xu, Bo Li, Guanglin Niu(Corresponding Author)
2025 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2025

• IFD is the first to propose a dual-stream identity-attention model that effectively compels the network to focus comprehensively on the regions containing distinctive identity information.

📃Paper

CCUP: A Controllable Synthetic Data Generation Pipeline for Pretraining Cloth-Changing Person Re-Identification Models
Yujian Zhao, Chengru Wu, Yinong Xu, Xuanzheng Du, Ruiyu Li, Guanglin Niu(Corresponding Author)
IEEE International Conference on Multimedia & Expo (ICME), 2025

• CCUP is a high-quality synthetic Cloth Changing Re-Identification (CC-ReID) dataset generated via a low-cost and controllable data generation pipeline, which is the first large-scale (over 1,000,000 images) dataset for the CC-ReID task.

📃Paper 💾Code

Tablebench: A comprehensive and complex benchmark for table question answering
Xianjie Wu, Jian Yang, Linzheng Chai, Ge Zhang, Jiaheng Liu, Xinrun Du, Di Liang, Daixin Shu, Xianfu Cheng, Tianzhen Sun, Guanglin Niu, Tongliang Li, Zhoujun Li
Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2025

• TableBench is a human-annotated comprehensive and complex TableQA benchmark comprising 886 samples across 18 fields, designed to facilitate fact-checking, numerical reasoning, data analysis, and visualization tasks.
• Our work is promoted by several media and forums, such as AINLP

🔧Project 📃Paper 💾Code

MdEval: Massively Multilingual Code Debugging
Shukai Liu, Linzheng Chai, Jian Yang, Jiajun Shi, He Zhu, Liran Wang, Ke Jin, Wei Zhang, Hualei Zhu, Shuyue Guo, Tao Sun, Jiaheng Liu, Yunlong Duan, Yu Hao, Liqun Yang, Guanglin Niu, Ge Zhang, Zhoujun Li

• MdEval is the first massively multilingual debugging benchmark, which includes 3.6K test samples of 18 programming languages and covers the automated program repair (APR) task, the code review (CR) task, and the bug identification (BI) task.

📃Paper

FuzzCoder: Byte-level Fuzzing Test via Large Language Model
Liqun Yang, Jian Yang, Chaoren Wei, Guanglin Niu, Ge Zhang, Yunli Wang, et al

• FuzzCoder formulates the fuzzing test as a sequenceto-sequence paradigm and then introduce the generation model to attack vulnerable positions by selecting proper mutation positions and strategies.

📃Paper 💾Code