This project investigates transformation mechanisms over property graphs to support causal inference. It aims to understand how graph-based data models can encode not only factual information but also causal relationships among entities.

The project is grounded in established theories of causality, including causal graphs, structural causal models and structural equations, which provide the conceptual basis. A central objective is to translate these theoretical constructs into first-order logic–based rules that can be applied to property graphs, enabling the identification, preservation, and controlled transformation of causal dependencies.

To achieve this, the project explores how expressive logical frameworks from data integration can be used to formalize causal constraints and inference rules over graphs. A key requirement is that graph transformations remain faithful to the underlying causal semantics and preserve essential properties for causal reasoning, such as d-separation, back-door criteria, and front-door criteria.