@inproceedings{zhong2025benchmarking,title={Benchmarking Retrieval-Augmented Generation for Chemistry},author={Zhong, Xianrui and Jin, Bowen and Ouyang, Siru and Shen, Yanzhen and Jin, Qiao and Fang, Yin and Lu, Zhiyong and Han, Jiawei},booktitle={Second Conference on Language Modeling},year={2025},url={https://openreview.net/forum?id=qG4dL0bart},}
NeurIPS
DynamicRAG: Leveraging Outputs of Large Language Model as Feedback for Dynamic Reranking in Retrieval-Augmented Generation
Jiashuo Sun, Xianrui Zhong, Sizhe Zhou, and 1 more author
In Thirty-ninth Conference on Neural Information Processing Systems , 2025
@inproceedings{sun2025dynamicrag,title={DynamicRAG: Leveraging Outputs of Large Language Model as Feedback for Dynamic Reranking in Retrieval-Augmented Generation},author={Sun, Jiashuo and Zhong, Xianrui and Zhou, Sizhe and Han, Jiawei},booktitle={Thirty-ninth Conference on Neural Information Processing Systems},year={2025},}
NeurIPS
SIMWORLD: An Open-ended Simulator for Agents in Physical and Social Worlds
Xiaokang Ye, Jiawei Ren, Yan Zhuang, and 13 more authors
In Thirty-ninth Conference on Neural Information Processing Systems , 2025
@inproceedings{ye2025simworld,title={SIMWORLD: An Open-ended Simulator for Agents in Physical and Social Worlds},author={Ye, Xiaokang and Ren, Jiawei and Zhuang, Yan and He, Xuhong and Liang, Yiming and Yang, Yiqing and Dogra, Mrinaal and Zhong, Xianrui and Liu, Eric and Benavente, Kevin and Nagaraju, Rajiv Mandya and Sharma, Dhruv Vivek and Ma, Ziqiao and Shu, Tianmin and Hu, Zhiting and Qin, Lianhui},booktitle={Thirty-ninth Conference on Neural Information Processing Systems},year={2025},}
NeurIPS
FGBench: A Dataset and Benchmark for Molecular Property Reasoning at Functional Group-Level in Large Language Models
Xuan Liu, Siru Ouyang, Xianrui Zhong, and 2 more authors
In Thirty-ninth Conference on Neural Information Processing Systems (Datasets and Benchmarks Track) , 2025
@inproceedings{liu2025fgbench,title={FGBench: A Dataset and Benchmark for Molecular Property Reasoning at Functional Group-Level in Large Language Models},author={Liu, Xuan and Ouyang, Siru and Zhong, Xianrui and Han, Jiawei and Zhao, Huimin},booktitle={Thirty-ninth Conference on Neural Information Processing Systems (Datasets and Benchmarks Track)},year={2025},}
2024
ACL
ActionIE: Action Extraction from Scientific Literature with Programming Languages
Xianrui Zhong*, Yufeng Du*, Siru Ouyang, and 6 more authors
In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , Aug 2024
Extraction of experimental procedures from human language in scientific literature and patents into actionable sequences in robotics language holds immense significance in scientific domains. Such an action extraction task is particularly challenging given the intricate details and context-dependent nature of the instructions, especially in fields like chemistry where reproducibility is paramount. In this paper, we introduce ActionIE, a method that leverages Large Language Models (LLMs) to bridge this divide by converting actions written in natural language into executable Python code. This enables us to capture the entities of interest, and the relationship between each action, given the features of Programming Languages. Utilizing linguistic cues identified by frequent patterns, ActionIE provides an improved mechanism to discern entities of interest. While our method is broadly applicable, we exemplify its power in the domain of chemical literature, wherein we focus on extracting experimental procedures for chemical synthesis. The code generated by our method can be easily transformed into robotics language which is in high demand in scientific fields. Comprehensive experiments demonstrate the superiority of our method. In addition, we propose a graph-based metric to more accurately reflect the precision of extraction. We also develop a dataset to address the scarcity of scientific literature occurred in existing datasets.
2023
EMNLP Demo
Reaction Miner: An Integrated System for Chemical Reaction Extraction from Textual Data
Ming Zhong, Siru Ouyang, Yizhu Jiao, and 15 more authors
In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: System Demonstrations , Dec 2023
Chemical reactions, as a core entity in the realm of chemistry, hold crucial implications in diverse areas ranging from hands-on laboratory research to advanced computational drug design. Despite a burgeoning interest in employing NLP techniques to extract these reactions, aligning this task with the real-world requirements of chemistry practitioners remains an ongoing challenge. In this paper, we present Reaction Miner, a system specifically designed to interact with raw scientific literature, delivering precise and more informative chemical reactions. Going beyond mere extraction, Reaction Miner integrates a holistic workflow: it accepts PDF files as input, bypassing the need for pre-processing and bolstering user accessibility. Subsequently, a text segmentation module ensures that the refined text encapsulates complete chemical reactions, augmenting the accuracy of extraction. Moreover, Reaction Miner broadens the scope of existing pre-defined reaction roles, including vital attributes previously neglected, thereby offering a more comprehensive depiction of chemical reactions. Evaluations conducted by chemistry domain users highlight the efficacy of each module in our system, demonstrating Reaction Miner as a powerful tool in this field.
@inproceedings{zhong-etal-2023-reaction,title={Reaction Miner: An Integrated System for Chemical Reaction Extraction from Textual Data},author={Zhong, Ming and Ouyang, Siru and Jiao, Yizhu and Kargupta, Priyanka and Luo, Leo and Shen, Yanzhen and Zhou, Bobby and Zhong, Xianrui and Liu, Xuan and Li, Hongxiang and Xiao, Jinfeng and Jiang, Minhao and Hu, Vivian and Wang, Xuan and Ji, Heng and Burke, Martin and Zhao, Huimin and Han, Jiawei},editor={Feng, Yansong and Lefever, Els},booktitle={Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: System Demonstrations},month=dec,year={2023},address={Singapore},publisher={Association for Computational Linguistics},}
2022
ICLR
Huber Additive Models for Non-stationary Time Series Analysis
Yingjie Wang, Xianrui Zhong, Fengxiang He, and 2 more authors
In International Conference on Learning Representations , Dec 2022
@inproceedings{wang2022huber,title={Huber Additive Models for Non-stationary Time Series Analysis},author={Wang, Yingjie and Zhong, Xianrui and He, Fengxiang and Chen, Hong and Tao, Dacheng},booktitle={International Conference on Learning Representations},year={2022},}
2021
PLOS One
Clinical determinants of the severity of COVID-19: A systematic review and meta-analysis
Xinyang Li, Xianrui Zhong, Yongbo Wang, and 3 more authors
Objective We aimed to systematically identify the possible risk factors responsible for severe cases. Methods We searched PubMed, Embase, Web of science and Cochrane Library for epidemiological studies of confirmed COVID-19, which include information about clinical characteristics and severity of patients’ disease. We analyzed the potential associations between clinical characteristics and severe cases. Results We identified a total of 41 eligible studies including 21060 patients with COVID-19. Severe cases were potentially associated with advanced age (Standard Mean Difference (SMD) = 1.73, 95% CI: 1.34–2.12), male gender (Odds Ratio (OR) = 1.51, 95% CI:1.33–1.71), obesity (OR = 1.89, 95% CI: 1.44–2.46), history of smoking (OR = 1.40, 95% CI:1.06–1.85), hypertension (OR = 2.42, 95% CI: 2.03–2.88), diabetes (OR = 2.40, 95% CI: 1.98–2.91), coronary heart disease (OR: 2.87, 95% CI: 2.22–3.71), chronic kidney disease (CKD) (OR = 2.97, 95% CI: 1.63–5.41), cerebrovascular disease (OR = 2.47, 95% CI: 1.54–3.97), chronic obstructive pulmonary disease (COPD) (OR = 2.88, 95% CI: 1.89–4.38), malignancy (OR = 2.60, 95% CI: 2.00–3.40), and chronic liver disease (OR = 1.51, 95% CI: 1.06–2.17). Acute respiratory distress syndrome (ARDS) (OR = 39.59, 95% CI: 19.99–78.41), shock (OR = 21.50, 95% CI: 10.49–44.06) and acute kidney injury (AKI) (OR = 8.84, 95% CI: 4.34–18.00) were most likely to prevent recovery. In summary, patients with severe conditions had a higher rate of comorbidities and complications than patients with non-severe conditions. Conclusion Patients who were male, with advanced age, obesity, a history of smoking, hypertension, diabetes, malignancy, coronary heart disease, hypertension, chronic liver disease, COPD, or CKD are more likely to develop severe COVID-19 symptoms. ARDS, shock and AKI were thought to be the main hinderances to recovery.
@article{10.1371/journal.pone.0250602,doi={10.1371/journal.pone.0250602},author={Li, Xinyang and Zhong, Xianrui and Wang, Yongbo and Zeng, Xiantao and Luo, Ting and Liu, Qing},journal={PLOS ONE},publisher={Public Library of Science},title={Clinical determinants of the severity of COVID-19: A systematic review and meta-analysis},year={2021},month=may,volume={16},pages={1-21},number={5},}