Reviewer, ACL Rolling Review (ARR)
2026 - Present

Undergraduate Student, Cyber Science and Engineering
Huazhong University of Science and Technology (HUST) · Qiming Experimental Class
normanspark@hust.edu.cn · zxinglang594@gmail.com
I am actively looking for 27 Spring Research Intern Opportunities and 28 Fall PhD Positions
Xinglang Zhang is a sophomore undergraduate majoring in Cyber Science and Engineering at Huazhong University of Science and Technology (HUST), advised by Professor Zikai Song at the Media Lab. His research focuses on neural-symbolic AI, computational social science and agent systems.
Xinglang is naturally proactive and collaborative—he enjoys initiating ideas, working with others, and turning abstract concepts into practical work. Beyond research, he values communication, responsibility, and long-term growth.
Still early in his academic journey, Xinglang is highly motivated, open to feedback, and always excited by meaningful challenges.
at ACL 2026!
, the only recipient from the School of Cyber Science and Engineering!
.Selected publications by Xinglang Zhang.
Semiotic Logical Hexagon Theory for LLM Logical Reasoning arXiv 2026
arXiv preprint, 2026. Extended from our ACL 2026 Oral paper.
Introduces HexLogicAgent and a formal logical hexagon theory that organizes complete semantic opposition before deduction, improving the reliability of LLM logical reasoning across multiple models and benchmarks.
Semantic-Aware Logical Reasoning via a Semiotic Framework ACL 2026 Oral
ACL 2026 Main Conference (Oral), 2026. Also on arXiv:2509.24765
Introduces LogicAgent, a multi-perspective reasoning framework grounded in semiotics, and RepublicQA, a philosophy-inspired benchmark.
Logical Phase Transitions: Understanding Collapse in LLM Logical Reasoning Accepted to ACL 2026
ACL 2026 Main Conference, 2026. Also on arXiv:2601.02902
Identifies critical logical complexity thresholds in LLMs and proposes a model-agnostic Logical Complexity Metric (LoCM) alongside a neuro-symbolic curriculum tuning framework.
Too Good to Be Real: AI Synthetic Preference Misreads What Real Humans Reward Under Review
Under review, 2026.
Proposed an ontology-based preference decomposition framework and uncovered a synthetic-real preference gap: LLMs overvalue explicit reasoning, while real platform users reward affective and expressive salience.
IntervenSim: Intervention-Aware Social Network Simulation for Opinion Dynamics
arXiv preprint, 2026. Also on arXiv:2604.06600
Introduces an intervention-aware closed-loop simulation framework that jointly models source-side interventions and crowd feedback to better capture evolving opinion dynamics in online social networks.
Social Intelligence Modeling: A Comprehensive Survey from Social Perception to Social Simulation Preprint 2026
Preprint, 2026.
Introduces Social Intelligence Modeling as a unified perspective spanning social perception, understanding, generation, and simulation.
EarthVerse: Benchmarking Scientific Agents Across Dynamic Earth Systems and Natural Hazards Preprint 2026
Preprint, 2026.
Benchmarks scientific agents on dynamic Earth systems and natural-hazard scenarios.
Reviewer, ACL Rolling Review (ARR)
2026 - Present
Xinglang Zhang's research interests include Neural-Symbolic AI, LLM Agents, and Computational Social Science.
Leader of English Association, Liuyang No.1 High School
September 2022 - September 2023
Class Representative, Qiming Experimental Class
September 2024 - Present
Head of Organization Department, School of Cyber Science and Engineering
September 2025 - Present
PyTorch, Transformers, LoRA, SFT, RLHF, vLLM, Deep Learning
I'm a huge fan of Taylor Swift and Lana Del Rey, and I enjoy collecting their vinyl albums.
Hachi from Chiikawa. He's cute, positive, hardworking, and kind to everyone around him.
Ultimate Frisbee.
I love taking photos for my friends, especially atmospheric portrait photography.