Ruochi Li, Jianzhe Lin, Haoxuan Zhang, Haihua Chen, Junhua Ding, Edward Gehringer, Yang Zhang

Paper  /  Homepage  /  Code  /  Dataset

A topology-aware benchmark for multimodal GraphRAG, assessing the retrieval and composition of heterogeneous, layout-grounded document evidence under controlled reasoning topologies.

Haoxuan Zhang, Ruochi Li, Yang Zhang, Zhenni Liang, Junhua Ding, Ting Xiao, Haihua Chen

Paper  /  Code

A large-scale benchmark for generative AI documentation, combining cross-source evidence triangulation, multi-agent curation, and human-in-the-loop validation to assess the faithfulness and completeness of model and data cards.

Haoxuan Zhang, Ruochi Li, Zhenni Liang, Mehri Sattari, Phat Vo, Collin Qu, Ting Xiao, Junhua Ding, Yang Zhang, Haihua Chen

Paper  /  Code

An adaptive retrieval-augmented framework that integrates context-aware query expansion with cross-card knowledge transfer to generate comprehensive, structured documentation for generative AI models and datasets.

Ruochi Li, Haoxuan Zhang, Edward Gehringer, Ting Xiao, Junhua Ding, Haihua Chen

Paper  /  Code

A systematic investigation of LLM-generated scientific peer reviews, combining semantic and graph-structural analyses to characterize their descriptive strengths, limitations in critical assessment, and sensitivity to paper quality.

Haoxuan Zhang, Ruochi Li, Sarthak Shrestha, Shree Harshini Mamidala, Revanth Putta, Arka Krishan Aggarwal, Ting Xiao, Junhua Ding, Haihua Chen

Paper

An LLM-driven framework for identifying deficient scientific peer reviews, combining human-validated annotation and targeted synthetic augmentation to strengthen automated review quality assessment.

Haoxuan Zhang, Ruochi Li, Yang Zhang, Ting Xiao, Jiangping Chen, Junhua Ding, Haihua Chen

Paper

A role-based synthesis of LLMs in scientific innovation, examining how autonomy, cognitive function, and human oversight shape their capabilities as research assistants, collaborators, scientists, and evaluators.

Yuhan Zhou, Ruochi Li, Ana Cleveland, Junhua Ding, Kewei Sha, Haihua Chen

Paper  /  Code

A quality-aware framework for medical concept normalization, integrating controlled LLM-based augmentation with multi-source data fusion to expand concept coverage and address long-tailed clinical terminology.

Haihua Chen, Yuhan Zhou, Ruochi Li, Aryan Murthy Illa, Ana Cleveland, Junhua Ding

Paper  /  Resources

A comprehensive review of medical concept normalization, organizing datasets, ontologies, modeling paradigms, and evaluation practices to clarify the field's methodological evolution and open challenges.