I'm looking for a collection of parsed mathematical formulas (expressions). I'm interested in studying the statistical properties of mathematical expressions written by humans. (Thus, formulas generated mechanically are not interesting.) For now, I'd prefer formulas restricted to the elementary functions.
I'm considering the following sources, all of which make available database dumps:
- Wikipedia articles
- StackExchange Q&As, particularly from MathOverflow, Math.SE, or Physics.SE
- arXiv papers about math and physics
These sources contain plenty of LaTeX formulas but they'd have to be extracted, parsed, and validated, which is by no means trivial.
Are there any datasets of naturally occurring mathematical formulas that would be easier to work with? Or, failing that, what other sources ought I consider?