## Project ideas, mainly ML

**BERT-based model for mathematics: **There exists a BERT-model trained on mathematical texts. However, it does not outperform models trained on broader scientific texts nor even on general BERT-models. Therefore, it is a project to scope existing SOTA models, define a corpus, train a model for mathematics, assess it, and apply it to an interesting question in philosophy of mathematics.

## Project ideas, mainly DH and PMP

**Narrative structure in mathematics: **Mathematical texts follow narrative structures both in terms of concepts and in framing and hedging. Therefore it is a project to develop ways of tracing internal structure of mathematical texts, for instance about the position and semantic roles of examples and counter examples. Such a framework would allow for philosophical analyses of the epistemic roles of such elements.

It is important to notice that these examples are **only suggestions** and that a real question and problem description will be developed in collaboration between student and supervisor.

The projects described above are deliberately broad; typically the project description phase would entail shortening and sharpening the question, leaving parts of the project suggestion out.

For instance, the above suggestions can be adapted to apply to e.g. computer science rather than mathematics as a discipline.

If you want to know more about past projects, please refer to our list of present and former group members.