Miss Su is a researcher specializing in trustworthy artificial intelligence, with a focus on natural language processing (NLP) and multimodal models. Her work tackles scientific misinformation, supports evidence-based climate policy, and enhances the robustness of intelligent systems. She leads the development of ClimateViz, a large-scale benchmark designed to evaluate the factual consistency and reasoning abilities of AI models in climate science. Her research integrates statistical reasoning, knowledge graph construction, and large language model evaluation to improve the transparency and reliability of AI. She is also exploring alignment and safety challenges in multimodal fact-checking, with applications in environmental science and global sustainability. Through interdisciplinary collaboration and cutting-edge methodology, her work advances AI systems that are both scientifically grounded and socially responsible.
Research Interests
Scientific fact-checking using large language and vision-language models (LLMs/LVLMs)
Multimodal reasoning over climate charts, tables, and text
Knowledge graph construction and query-based evaluation for factual consistency
Evaluation and alignment of AI models in scientific domains
Synthetic data generation for scalable model training and benchmarking
Interpretability and safety in multimodal fact-checking systems
NLP for climate policy and environmental communication
AI-driven decision support for sustainability and intelligent networks
