The University of Southern Maine’s Department of Computer Science continues to expand its international research presence as Nathaniel Serrano, a member of the USM Artificial Intelligence and Information Retrieval Laboratory, presented research at the ACM SIGIR 2026 Conference on Research and Development in Information Retrieval in Melbourne, Australia.
ACM SIGIR is one of the world’s premier conferences in information retrieval, bringing together researchers, students, and industry professionals to share advances in search technologies, artificial intelligence, machine learning, and information access. Presenting research at SIGIR provides an opportunity for researchers to engage with the global research community and share innovative approaches to improving how people interact with digital information.
At the conference, Serrano presented a poster on the paper, “MathMex-PDF: Towards Accessible Visual Mathematics,” which introduces a new approach for searching and interacting with mathematical content contained within PDF documents.
Mathematical information retrieval remains challenging because readers often use standard PDF viewers for math-heavy documents while relying on separate tools for searching mathematical expressions. MathMex-PDF addresses this challenge by providing an interactive PDF reader that combines mathematical expression extraction and multimodal retrieval within a single workflow.
The system processes raw PDF pages to detect and recognize mathematical expressions, grounds those expressions within the document layout, and enables unified search across both text and formulas. Users can submit natural-language, LaTeX, or hybrid queries, allowing them to search mathematical documents using multiple forms of input. The system also supports interactive features such as click-to-search reverse formula lookup, enabling users to explore mathematical content directly from rendered pages.
The research was conducted by members of the USM Artificial Intelligence and Information Retrieval Laboratory, including Nathaniel Serrano, Lucas Matheson, Clayton Durepos, and Dr. Behrooz Mansouri. The project contributes to ongoing efforts to improve information retrieval systems and make complex digital content more accessible through artificial intelligence.
Paper: MathMex-PDF: Towards Accessible Visual Mathematics
DOI: https://dl.acm.org/doi/10.1145/3805712.3808386