Causality by Abstraction: Symbolic Rule Learning in Multivariate Timeseries with Large Language Models

The project introduces a method for symbolic abstraction of time-series data using LLMs. While scientifically interesting, it currently exists as a low-traction research artifact (0 stars) with no clear moat beyond the methodology described in the associated paper. Frontier labs are increasingly moving into reasoning and time-series analysis, making this specific niche vulnerable to general-purpose model capabilities in the medium term.