MioFFAn: Open-source tool uses LLMs to turn math formulas into executable code

The paper titled "MioFFAn: an Annotation Software for Formula Formalization with LLM Automation Capabilities" introduces a new open-source framework designed to tackle the challenge of automatically translating mathematical expressions from scientific literature into executable symbolic code (a process called Formula Formalization). The authors—Nicolas Sibuet, Horacio Saggion, and Riccardo Rossi—identify a severe scarcity of high-quality ground-truth datasets specialized for technical scientific domains as a key bottleneck.

MioFFAn builds upon the MioGatto architecture, extending its features to support custom taxonomies, selection of equations of interest, and aided symbolic code specification. Critically, it incorporates partial automation via large language models by defining a modular set of sub-tasks with strict output formats. This allows researchers to iteratively refine automation strategies and evaluate them using standard NLP metrics. The preliminary evaluation demonstrates the effectiveness of this human-in-the-loop approach. The work was presented at the 3rd International Workshop on Natural Scientific Language Processing (NSLP 2026), co-located with LREC2026.