About Mathoscope

Mathoscope (formerly Parmesan) is an experimental search system for mathematical concepts, specifically in category theory. The system was created in 2023 by Jacob Collard, Valeria de Paiva, and Eswaran Subrahmanian. The basic idea behind it is that a newcomer to the field may come across unfamiliar concepts and wish to find out more about them. This isn't limited to just learning their definitions or proofs, but also includes finding examples that highlight how they are used in real mathematical texts.

Mathoscope has been presented at three scientific conferences: W-NUT 2022, 4LR 2023, and LREC-COLING 2024. Correspondingly, we have two preprints describing the functionality of Mathoscope as well as our broader findings on working with mathematical text as well as a paper in the LREC-COLING proceedings. All of these are available on arXiv.

• Jacob Collard, Valeria de Paiva, Brendan Fong, and Eswaran Subrahmanian. 2022. Extracting Mathematical Concepts from Text. ArXiv preprint. 2208.13830.
• Jacob Collard, Valeria de Paiva, Eswaran Subrahmanian. 2023. Parmesan: mathematical concept extraction for education. ArXiv preprint. 2307.06699.
• Jacob Collard, Valeria de Paiva, Eswaran Subrahmanian. 2024. Mathematical Entities: Corpora and Benchmarks. Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation. LREC-COLING

The presentation at W-NUT 2022 is available as a recorded video, presented by Jacob Collard.

The presentation at LREC-COLING is also available in video form, presented by Jacob Collard.