Causality by Abstraction: Symbolic Rule Learning in Multivariate Timeseries with Large Language Models
Preetom Biswas, Giulia Pedrielli, K. Selçuk Candan · 2026
Read Causality by Abstraction: Symbolic Rule Learning in Multivariate Timeseries with Large Language Models by Preetom Biswas, Giulia Pedrielli, K. Selçuk Candan (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-17829.txt.
Description
Preetom Biswas and team present ruleXplain, a frame that uses large language models to extract formal causal links in simulation-driven motion systems. Inferring causal links in time series with delayed effects is hard, above all when complex motion defies plain maps. Standard methods often fail to give generalized, plain explanations, since distinct input paths can yield near-identical outputs.
The method introduces a constrained sign-based rule language with time tools and delay meanings. Language models generate rules open to checks through guided prompts, fed by grouped what-if input paths from a simulator that yield similar target outputs. A closed-loop refinement process ensures rule match and meaning reach. Tests use the PySIRTEM outbreak simulator, mapping testing rates to daily case counts, and the EnergyPlus building simulator, mapping heat and solar inputs to power needs. Three test classes check rule power, causal coding and carry-over.
Catalog record
| CREATOR | Preetom Biswas, Giulia Pedrielli, K. Selçuk Candan |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 58 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-17829