Bohan Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Bohan Li has a total of 241 co-authors. Here are the top co-authors by number of shared publications:
Anman Zhang
Nanjing University of Aeronautics and Astronautics
7 shared publications- Ken Cai
College of Automation, Zhongkai University of Agriculture and Engineering, Guangzhou, China
7 shared publications - Haofen Wang
College of Design and Innovation, Tongji University, Shanghai, China
6 shared publications - Hongzhi Yin
The University of Queensland
6 shared publications Shuo Wan
Nanjing University of Aeronautics and Astronautics
6 shared publicationsDonghai Guan
5 shared publications
Latest Publications
- Open Access
- 6 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 4786-4795
Transcending the Semantic Chasm: Knowledge-Graph Guided Aggregation for Cross-Silo Federated Session Recommendation
- 6 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 4786-4795
Open AccessTranscending the Semantic Chasm: Knowledge-Graph Guided Aggregation for Cross-Silo Federated Session Recommendation
- 12 July 2026
- Published by Springer Nature
- p. 157-166
EvidITI: Enhancing RAG Robustness Through Inference-Time Intervention on Evidence-Seeking Attention Heads
- 12 July 2026
- Published by Springer Nature
- p. 157-166
EvidITI: Enhancing RAG Robustness Through Inference-Time Intervention on Evidence-Seeking Attention Heads
- Top Cited PapersOpen Access
- 16 February 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Information Systems
- Vol. 44 (3) , 1-30
U-NIAH: Unified RAG and LLM Evaluation for Long Context Needle-in-a-Haystack
Recent advancements in Large Language Models (LLMs) have significantly extended context windows, igniting discussions about the necessity of Retrieval-Augmented Generation (RAG). U-NIAH, a unified Needle-in-a-Haystack (NIAH) framework, systematically evaluates LLMs and RAG methods in controlled long-context settings. It extends beyond traditional NIAH by incorporating more practical and complex scenarios like multi-needle, long-needle, and needle-in-needle configurations and leveraging the synthetic dataset to mitigate LLM biases. The experiments aim to address three research questions in long-context scenarios: (1) performance tradeoffs between LLMs and RAG, (2) error patterns in RAG, and (3) RAG’s limitations in complex settings. Results show that smaller LLMs benefit more from RAG. In all settings, RAG achieves a win rate of 82.58% over direct answers. Additionally, it is found that retrieval noise and chunk ordering degrade RAG performance, and we further summarized typical error patterns, including omissions due to noise, hallucinations under high noise critical conditions, and self-doubt behaviors, as well as how these phenomena vary with context length. Finally, in some challenging scenarios, experiments show that deep reasoning models are more easily affected by distractors. These findings highlight the complementary roles of RAG and LLMs and offer actionable insights for optimizing deployment strategies ( https://github.com/Tongji-KGLLM/U-NIAH ).
- 16 February 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Information Systems
- Vol. 44 (3) , 1-30
Top Cited PapersOpen AccessU-NIAH: Unified RAG and LLM Evaluation for Long Context Needle-in-a-Haystack
Recent advancements in Large Language Models (LLMs) have significantly extended context windows, igniting discussions about the necessity of Retrieval-Augmented Generation (RAG). U-NIAH, a unified Needle-in-a-Haystack (NIAH) framework, systematically evaluates LLMs and RAG methods in controlled long-context settings. It extends beyond traditional NIAH by incorporating more practical and complex scenarios like multi-needle, long-needle, and needle-in-needle configurations and leveraging the synthetic dataset to mitigate LLM biases. The experiments aim to address three research questions in long-context scenarios: (1) performance tradeoffs between LLMs and RAG, (2) error patterns in RAG, and (3) RAG’s limitations in complex settings. Results show that smaller LLMs benefit more from RAG. In all settings, RAG achieves a win rate of 82.58% over direct answers. Additionally, it is found that retrieval noise and chunk ordering degrade RAG performance, and we further summarized typical error patterns, including omissions due to noise, hallucinations under high noise critical conditions, and self-doubt behaviors, as well as how these phenomena vary with context length. Finally, in some challenging scenarios, experiments show that deep reasoning models are more easily affected by distractors. These findings highlight the complementary roles of RAG and LLMs and offer actionable insights for optimizing deployment strategies ( https://github.com/Tongji-KGLLM/U-NIAH ).
- 2 February 2026
- Published by Springer Nature
- p. 357-368
FasterTune: A New Paradigm for Database Tuning with Large Language Models and Bayesian Optimization
- 2 February 2026
- Published by Springer Nature
- p. 357-368
FasterTune: A New Paradigm for Database Tuning with Large Language Models and Bayesian Optimization
- Open Access
- 26 January 2026
- Published by Springer Nature in Data Science and Engineering
- Vol. 11 (1) , 213-229
PPA++: Preference Prototype-Aware Learning with Large Language Model for Universal Cross-Domain Recommendation
- 26 January 2026
- Published by Springer Nature in Data Science and Engineering
- Vol. 11 (1) , 213-229
Open AccessPPA++: Preference Prototype-Aware Learning with Large Language Model for Universal Cross-Domain Recommendation
- 3 January 2026
- Published by Springer Nature
- p. 151-167
Self-supervised Dual Graph and Intention Association for Session-Based Recommendation
- 3 January 2026
- Published by Springer Nature
- p. 151-167
Self-supervised Dual Graph and Intention Association for Session-Based Recommendation
- 2 January 2026
- Published by Springer Nature
- p. 703-713
Grayscale Image-Based Top-k Spatial Dataset Search Processing
- 2 January 2026
- Published by Springer Nature
- p. 703-713
Grayscale Image-Based Top-k Spatial Dataset Search Processing
- 2 January 2026
- Published by Springer Nature
- p. 305-320
GoT-R: Enhancing Large Language Models for Complex Question Answering with Graph-of-Thought Guided Reasoning
- 2 January 2026
- Published by Springer Nature
- p. 305-320
GoT-R: Enhancing Large Language Models for Complex Question Answering with Graph-of-Thought Guided Reasoning
- Open Access
- 10 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 373-382
NR-GCF: Graph Collaborative Filtering with Improved Noise Resistance
- 10 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 373-382
Open AccessNR-GCF: Graph Collaborative Filtering with Improved Noise Resistance
- Open Access
An Efficient Proximity Graph-based Approach to Table Union Search
Neural embedding models are extensively employed in the table union search problem, which aims to find semantically compatible tables that can be merged with a given query table. In particular, multi-vector models, which represent a table as a vector set (typically one vector per column), have been demonstrated to achieve superior retrieval quality by capturing fine-grained semantic alignments. However, this problem faces more severe efficiency challenges than the single-vector problem due to the inherent dependency on bipartite graph maximum matching to compute unionability scores. Therefore, this paper proposes an efficient Proximity Graph-based Table Union Search (PGTUS) approach. PGTUS employs a multi-stage pipeline that combines a novel refinement strategy, a filtering strategy based on many-to-one bipartite matching. Besides, we propose an enhanced pruning strategy to prune the candidate set, which further improve the search efficiency. Extensive experiments on six benchmark datasets demonstrate that our approach achieves 3.6-6.0X speedup over existing approaches while maintaining comparable recall rates.
Open AccessAn Efficient Proximity Graph-based Approach to Table Union Search
Neural embedding models are extensively employed in the table union search problem, which aims to find semantically compatible tables that can be merged with a given query table. In particular, multi-vector models, which represent a table as a vector set (typically one vector per column), have been demonstrated to achieve superior retrieval quality by capturing fine-grained semantic alignments. However, this problem faces more severe efficiency challenges than the single-vector problem due to the inherent dependency on bipartite graph maximum matching to compute unionability scores. Therefore, this paper proposes an efficient Proximity Graph-based Table Union Search (PGTUS) approach. PGTUS employs a multi-stage pipeline that combines a novel refinement strategy, a filtering strategy based on many-to-one bipartite matching. Besides, we propose an enhanced pruning strategy to prune the candidate set, which further improve the search efficiency. Extensive experiments on six benchmark datasets demonstrate that our approach achieves 3.6-6.0X speedup over existing approaches while maintaining comparable recall rates.