Alt Text for Academic Figures
2026A vision-language pipeline that generates descriptive alt text for figures in academic manuscripts, improving accessibility of scholarly content for screen-reader users.
Research integrity · applied ML
Catching fraudulent papers and rethinking academic processes in the age of AI empowered Humans.
Catching fraudulent papers, AI-generated content, and citation manipulation—and building tools like Cite Lens to support editorial screening.
Embedding-based recommendation and retrieval over the scholarly literature, from domain-specific pretraining to reviewer and journal matching.
Deploying and serving on-premise LLMs at scale—up to 72B parameters—powering RAG pipelines and internal tooling company-wide.
A vision-language pipeline that generates descriptive alt text for figures in academic manuscripts, improving accessibility of scholarly content for screen-reader users.
MDPI's in-house research integrity ecosystem: AI-driven detection of paper mills, AI-generated content, image manipulation, and citation/author manipulation, paired with human-in-the-loop editorial review.
Book chapter
Introduces Cite Lens, an embedding-similarity tool for detecting problematic citation behavior—either misalignment between an article and its references, or between a reference and its citing paragraph. We analyze citation patterns across publishers and topics and show how this approach supports editorial screening against unethical or manipulative citations.