AI for Science
Scientific data and knowledge intelligence for literature mining, scientific modeling, reasoning, and autonomous discovery.
Mingyue Cheng 程明月
Ph.D. · Associate Researcher
State Key Laboratory of Cognitive Intelligence
School of Computer Science and Technology, University of Science and Technology of China
School of Computer Science and Technology, University of Science and Technology of China
I build prediction intelligence for complex systems by combining time-series observations, scientific knowledge, and agentic reasoning.
Email: mycheng@ustc.edu.cn
Address: B709, Xinzhi Building, Gaoxin Campus of USTC, Hefei, Anhui, China, 230031
Address: B709, Xinzhi Building, Gaoxin Campus of USTC, Hefei, Anhui, China, 230031
Biography
I am an Associate Researcher at the School of Computer Science and Technology, University of Science and Technology of China (USTC). I am also affiliated with the State Key Laboratory of Cognitive Intelligence at USTC, working under the guidance of Prof. Enhong Chen and Prof. Qi Liu. Previously, I obtained my Ph.D. degree under the supervision of Prof. Qi Liu.
Research Interests
My research develops cognitive intelligence methods for complex data mining, centered on LLMs and Agentic AI, and driven by the dual foundations of time-series observations and scientific knowledge. My methodological focus lies in context representation and reasoning, aiming to build predictive intelligence for complex systems through multimodal semantic understanding, slow-thinking temporal reasoning, and autonomous agentic interaction.
- 🤖LLMs and Agentic AI: Developing autonomous interactive learning for large language models, including environment-interactive Agentic RL, tool-augmented reasoning, multi-agent orchestration, and continual capability evolution through context, knowledge, and memory.
- 📊Time Series Intelligence: Developing context-aware predictive intelligence, with a focus on multimodal context representation, slow-thinking reasoning, uncertainty-aware forecasting, and autonomous agentic interaction.
- 📚 Scientific Knowledge Cognition: Developing agent-ready scientific knowledge intelligence through LLM-driven literature mining, multimodal scientific document understanding, evidence-grounded reasoning, and scientific knowledge integration for predictive modeling and decision support.
Application Domains and Evaluation Scenarios
Industrial Systems
Predictive intelligence for real-world complex systems, including energy, traffic, cloud services, finance, and industrial operations.
Recommender Systems
Adaptive user intelligence and personalized recommendation through behavior understanding, preference modeling, and contextual reasoning.
Prospective students and research collaborators are welcome to explore the
USTC-AGI Group
and contact me by email.
Latest News
- [Aug. 2026] 🎉 Congratulations on our paper PaperScout being accepted to Findings of EMNLP 2026!
- [Aug. 2026] 🎉 We are excited to present our tutorial, “Context-Aware Time Series Forecasting: From Pattern Extrapolation to Cognitive Reasoning,” at IEEE ICDM 2026. See you in Shenyang!
- [Aug. 2026] 🎉 I will serve as the Publication Chair for ICEBE 2026.
- [Aug. 2026] 🎉 Congratulations on our papers Mind2Report, Time-R1, and AlphaCast being accepted to ACM CIKM 2026!
- [Aug. 2026] 🎉 Congratulations on our survey A Survey on Table Mining with Large Language Models: Challenges, Advancements and Prospects being accepted by ACM Computing Surveys (ACM CSUR)!
- [Aug. 2026] 🎉 Congratulations on our demo papers Agent-R1 and TabClaw being accepted to the ACM CIKM 2026 Demo Track!
- [Jul. 2026] Our tutorial on Context-Aware Time Series Forecasting has been accepted by the IEEE ICDM 2026 Industry Tutorial Session. I will serve as a tutorial organizer and speaker.
- [Jul. 2026] Our workshop, FedKDD/FedMAS 2026, will be held in conjunction with KDD 2026. I am honored to serve as the Program Chair.
- [Jul. 2026] 🏆 Our Agentic Time Series Forecasting solution CastStar ranked #1 on the GIFT-Eval Overall leaderboard.
- [Jun. 2026] 🎉 Our project CastClaw（观星阁） received a Poster Presentation Certificate at the 2026 BAAI Conference Agent for Science Competition.
- [Jun. 2026] 🎉 Congratulations on our Knowledge-Oriented RAG Survey being accepted to ACM Transactions on Information Systems (ACM TOIS)!
- [May. 2026] 🎉 Congratulations to Zhiding Liu on his work DisenTS being accepted to IEEE Transactions on Pattern Analysis and Machine Intelligence (IEEE TPAMI)!
- [May. 2026] 🎉 Congratulations to Xiaoyu Tao on her paper being accepted to ACM Transactions on Intelligent Systems and Technology (ACM TIST)!
- [May. 2026] 🎉 Congratulations to two of my co-supervised students (Yucong Luo and Jie Ouyang) on receiving ByteDance Jindouyun offers!
- [May. 2026] 🎉 Congratulations to Yitong Zhou on our paper ChemTable being accepted to KDD 2026!
- [May. 2026] 🚀 We released CastFactory（炼星坊）, a framework for LLM-driven time series forecasting model training. GitHub
- [May. 2026] 🎉 Congratulations to Xiaoyu Tao and Bohou Zhang on our papers (TokenCast and ScholarSum) being accepted to IJCAI 2026!
- [May. 2026] 🎉 Congratulations to Yaguo Liu and Xiaoyu Tao on our papers (CoGenCast and MemCast) being accepted to ICML 2026!
- [Apr. 2026] 📄 We released the preprint version of our survey on LLM-based agents from the contextual cognition perspective on Preprints.org: "A Comprehensive Survey of the LLM-Based Agent: The Contextual Cognition Perspective".
- Mingyue Cheng, Daoyu Wang, Qi Liu*, Shuo Yu, Xiaoyu Tao, Yuqian Wang, Chengzhong Chu, Yu Duan, Mingkang Long, Enhong Chen, Mind2Report: A Cognitive Deep Research Agent for Expert-Level Commercial Report Synthesis. ACM CIKM 2026 Accepted. [PDF] [Code]
- Yitong Zhou, Yucong Luo, Mingyue Cheng*, Jiahao Wang, Daoyu Wang, Tingyue Pan, Jintao Zhang, Qi Liu, Enhong Chen, Time Series Forecasting as Reasoning: A Slow-Thinking Approach with Reinforced LLMs. ACM CIKM 2026 Accepted. [PDF] [Code]
- Xiaohan Zhang, Tian Gao, Mingyue Cheng*, Bokai Pan, Ze Guo, Yaguo Liu, Xiaoyu Tao, Qi Liu, AlphaCast: A Human Wisdom-LLM Intelligence Co-Reasoning Framework for Interactive Time Series Forecasting. ACM CIKM 2026 Accepted. [PDF] [Code]
- Tingyue Pan, Jie Ouyang, Mingyue Cheng, Qingchuan Li, Zirui Liu, Daoyu Wang, Mingfan Pan, Shuo Yu, Qi Liu, PaperScout: An Autonomous Agent for Academic Paper Search with Process-Aware Sequence-Level Policy Optimization. Findings of EMNLP 2026 Accepted. [PDF]
- Mingyue Cheng, Qingyang Mao, Qi Liu*, Yitong Zhou, Yupeng Li, Jiahao Wang, Jiaying Lin, Jiawei Cao, Enhong Chen, A Survey on Table Mining with Large Language Models: Challenges, Advancements and Prospects. ACM Computing Surveys (ACM CSUR) Accepted. [PDF] [Code]
- Tingyue Pan, Mingyue Cheng*, Shilong Zhang, Zhiding Liu, Xiaoyu Tao, Yucong Luo, Jintao Zhang, Qi Liu, OneCast: Structured Decomposition and Modular Generation for Cross-Domain Time Series Forecasting. ACM Transactions on Knowledge Discovery from Data (ACM TKDD) Accepted. [PDF]
- Xiaoyu Tao, Mingyue Cheng*, Ze Guo, Shuo Yu, Yaguo Liu, Qi Liu, Shijin Wang, MemCast: Memory-Driven Time Series Forecasting with Experience-Conditioned Reasoning. ICML2026 Accepted. [PDF]
- Mingyue Cheng, Jiahao Wang, Daoyu Wang, Xiaoyu Tao, Qi Liu*, Enhong Chen, Can Slow-Thinking LLMs Reason Over Time? Empirical Studies in Time Series Forecasting. ACM WSDM2026: 99–110, Feb 2026. [PDF] [Code]
Open Source
CastClaw（观星阁）
CastClaw（观星阁） is an open-source framework for agentic time series forecasting. It enables LLM agents to autonomously perform data analysis, model selection, feature engineering, and iterative forecasting refinement through structured tool-augmented workflows. CastClaw（观星阁） bridges classical time series methods with agentic reasoning, supporting diverse forecasting scenarios ranging from short-term point prediction to long-horizon multi-step forecasting, with built-in evaluation and interpretability mechanisms.
CastFactory（炼星坊）
CastFactory（炼星坊） is an open-source framework for LLM-driven time series forecasting model training, designed to make it extremely easy for users to build and adapt TSF models with modern large-model pipelines. It provides a unified and practical workflow for continued pre-training (CPT), supervised fine-tuning (SFT), and reinforcement learning (RL), helping researchers and practitioners develop, optimize, and evaluate LLM-based forecasting systems with much lower engineering overhead.
FutureCast（天星台）
FutureCast（天星台） is a unified evaluation suite for time series models and agentic forecasting systems. It organizes benchmark management, metric computation, error attribution, and report generation into a consistent evaluation workflow, helping researchers compare model capabilities and system-level gains across datasets, scenarios, and task settings including forecasting, classification, and anomaly detection.
NeoResearch（智多星）
NeoResearch（智多星） is an autonomous research agent system for time series forecasting model development. It connects research hypothesis generation, candidate recipe search, controlled experiments, evaluation diagnosis, and research memory into a reproducible loop, supporting systematic iteration from problem definition to validated forecasting model candidates.
Academic Search
Academic Search is an open-source academic literature research skill for Claude Code. It unifies paper discovery across arXiv, Semantic Scholar, Google Scholar, CNKI, and other scholarly sources, while supporting query expansion, citation tracing, BibTeX export, PDF-first retrieval, and multi-source deduplication. With recency-aware ranking and browser-assisted fallback for hard-to-access platforms, it helps researchers surface recent, high-value papers and their code resources more efficiently.
PaperScout
PaperScout is an autonomous academic paper search agent trained with Process-Aware Sequence-Level Policy Optimization (PSPO). It formulates literature search as a multi-turn sequential decision-making problem, allowing agents to dynamically decide when to Search for new directions and when to Expand along citation and related-paper paths. PaperScout turns scholarly retrieval into an agentic RL environment for learning efficient tool-use policies and producing stronger recall under realistic research queries.
TabClaw
TabClaw is an open-source agentic framework that empowers LLMs to reason over complex, real-world tabular data. It decomposes table-centric tasks into structured sub-goals, equips agents with code execution, schema-aware lookup, and formula tools, and coordinates them through multi-step decision workflows. TabClaw is designed to tackle challenges beyond flat QA — including multi-hop joins, conditional aggregation, and cross-table inference — making it suitable for enterprise data analysis and scientific table understanding.
Claw-R1
Claw-R1 is an open-source framework for training reasoning-intensive, tool-using LLM agents via reinforcement learning. It extends standard RL environments with agentic action spaces — including tool invocation, multi-turn interaction, and environment feedback — and supports GRPO-based policy optimization with customizable reward signals. Designed for reproducibility and extensibility, Claw-R1 enables researchers to study how RL shapes emergent reasoning behaviors, tool-use strategies, and decision robustness in large language models.
Agent-R1
Agent-R1 is a large-model agent training framework for end-to-end reinforcement learning fine-tuning. It supports GRPO-based policy optimization with reward signals derived from environment feedback, and equips agents with multi-tool orchestration, persistent long-term memory, and reflective self-correction across interaction turns. Grounded in the DeepSeek-R1 paradigm, Agent-R1 enables researchers to train LLM agents that not only reason under uncertainty but also learn to use tools strategically and recover from failures autonomously.
Benchmarks & Datasets
E-Commerce Search · Recall to Relevance
KuaiSearch: A Large-Scale E-Commerce Search Dataset for Recall, Ranking, and Relevance is built from real user search interactions on Kuaishou. It preserves authentic user queries and natural-language product texts, covers
cold-start users and long-tail products, and spans the three key stages of modern search systems: recall, ranking, and relevance judgment.
Scientific Literature · Agentic Evaluation
PaperArena: An Evaluation Benchmark for Tool-Augmented Agentic Reasoning on Scientific Literature benchmarks agentic systems on scientific reading and reasoning with tool use. It targets
literature understanding, tool-augmented reasoning, and evidence-grounded evaluation over scientific papers and related scholarly workflows.
AI for Science · Chemical Tables
Benchmarking Multimodal LLMs on Recognition and Understanding over Chemical Tables introduces ChemTable, a real-world benchmark curated from chemical literature. It supports two core tasks:
table recognition and table understanding, with expert annotations over cell polygons, logical layouts, and chemistry-specific semantic labels such as reagents, catalysts, yields, and graphical components.
RAG Evaluation · Dynamic Benchmark
HoH: A Dynamic Benchmark for Evaluating the Impact of Outdated Information on RAG studies how retrieval-augmented generation systems fail when knowledge becomes stale. It provides a dynamic evaluation setting for measuring
temporal robustness, outdated-information sensitivity, and the impact of knowledge freshness in modern RAG pipelines.
Education
- Sep. 2017 ~ Jun. 2023, Major: Computer Science and Technology, Ph.D. degree, University of Science and Technology of China (USTC), China
- Sep. 2013 ~ Jul. 2017, Major: Marketing Management, Bachelor of Management, Hefei University of Technology (HFUT), China
Professional Experience
- Jun. 2023 ~ Present, Associate Researcher, School of Computer Science and Technology, USTC, Hefei, China
- Jan. 2021 ~ Jun. 2023, Research Assistant, Research Center for Big Data in Intelligent Healthcare, USTC, Hefei, China
- Jun. 2019 ~ Oct. 2019, Research Intern, Data Service Center (DSC), Platform and Content Group (PCG), Tencent, Shenzhen, China
- Jul. 2018 ~ Sep. 2018, Engineer Intern, Lab of Big Data, Core Technology R&D Platform, iFLYTEK, Hefei, China
Teaching
- Introduction to Data Science（数据分析及实践）, Undergraduate Course, Spring 2024.
- Introduction to Data Science（数据分析及实践）, Undergraduate Course, Spring 2025.
- Introduction to Data Science（数据分析及实践）, Undergraduate Course, Spring 2026.
- Introduction to Systems Programming（系统程序设计基础）, Undergraduate Course, Spring 2026.
Honors and Awards
- Aug. 2025, 中国科学技术大学红专青年人才
- Mar. 2025, Best of WSDM2025（优秀论文奖）
- Dec. 2024, 昇腾人工智能创新大赛全国总决赛金奖（指导教师）
- Aug. 2024, Second Place of Task2 & Task3 in KDD Cup'2024
- Jan. 2024, 中国科学技术大学雄鹰创新创业基金（指导教师）
- Dec. 2023, 2023年中国国际大学生创新大赛国家级银奖（教育部）
- Jul. 2023, 全国"互联网+"大学生创新创业大赛安徽省金奖（安徽省教育厅）
- Mar. 2023, 中国科学技术大学优秀毕业生
- May. 2022, 2022年中国科学技术大学"创新创业"基金
- Dec. 2021, "创青春"中国青年创新创业项目支持——大学生创业"金种子"计划（团中央）
- Sep. 2021, 博士研究生国家奖学金（教育部）
- Sep. 2019, 腾讯科技平台与事业群实习生Mini项目卓越奖
Professional Service
Program Committee Member
- AAAI Conference on Artificial Intelligence (AAAI): 2026, 2027
- ACM International World Wide Web Conference (TheWebConf): 2024, 2025, 2026
- ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD): 2023, 2024, 2025, 2026, 2027
- ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (SIGSPATIAL): 2026
- International Conference on Learning Representations (ICLR): 2025
- International Conference on Machine Learning (ICML): 2025, 2026
- Conference on Neural Information Processing Systems (NeurIPS): 2024, 2025, 2026
- Annual Meeting of the Association for Computational Linguistics (ACL): 2025, 2026
- International Joint Conference on Artificial Intelligence (IJCAI): 2024, 2025, 2026 (SPC)
- International Conference on Web Search and Data Mining (WSDM): 2025
- SIAM International Conference on Data Mining (SDM): 2024
- ACM International Conference on Information and Knowledge Management (CIKM): 2024, 2025
- Database Systems for Advanced Applications (DASFAA): 2024, 2025
- IEEE Task Force on Data-Efficient Agentic Learning (DEAL)
- IEEE Task Force on AI for Time Series and Spatio-Temporal Data
Journal Reviewer
- IEEE Transactions on Knowledge and Data Engineering (IEEE TKDE)
- IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)
- IEEE Transactions on Neural Network and Learning Systems (IEEE TNNLS)
- IEEE Transactions on Industrial Informatics (IEEE TII)
- IEEE Transactions on Audio, Speech and Language Processing (IEEE T-ASL)
- ACM Transactions on Knowledge Discovery from Data (TKDD)
- ACM Transactions on Intelligent Systems and Technology (TIST)
- Transactions on Machine Learning Research (TMLR)
- Neurocomputing
- Frontiers of Computer Science (FCS)
- 软件学报
- 计算机学报
Research Grants
- 2026.08–2028.07, New Generation Artificial Intelligence–National Science and Technology Major Project
- 2026.08–2029.07, the Strategic Priority Research Program (B) of the Chinese Academy of Sciences
- 2026.01–2028.12, the Natural Science Foundation of China
- 2025.01–2027.12, USTC New Medicine Joint Fund Cultivation Project (Double First-Class Initiative Special Project)
- 2024.09–2026.08, Anhui Provincial Natural Science Foundation
- In addition, my research has been partially supported by industrial grants from leading companies such as Huawei, Kuaishou, and Tencent.