ERC GO-Y Bonifati (ERC GO-Y Bonifati)

Unifying Graph Database and Causal Models.

Graphs are expressive data structures representing more effectively relationships in data and enabling complex data-intensive tasks.
Graphs are also a widely adopted paradigm in causal inference focusing on causal directed acyclic graphs. Causal DAGs (Directed Acyclic Graphs) are manually curated by domain experts, but they are not validated, stored, integrated and versioned as data artifacts in a graph data system. In the GOY (pronounced GO Why) project, 1 focus on a radical shift towards causality-driven graph databases, addressing the development of solid theoretical foundations and a set of adequate tools underpinning causal graph operations. To date, these two paradigıns, namely causality and graph data management, are developed by entirely separate communities with different motivations: one is interested in causal analysis and inference, realized by mcans of ad-hoc programs and scripts on empirically validated causal graphs, while the other is interested in collecting, validating, integrating and querying graph data by means of declarative languages. These two areas are investigated in isolation although combining them creates exciting possibilities. The present proposal aims at investigating these possibilities and focuses on laying the theoretical foundations of causality-driven graph data management, a completely novel approach to data management guided by the theory of causation. Causal graphs, either provided by domain experts or extracted from the underlying observational data, are further semantically enriched. disambiguated, reconciled and used in the downstream processes of the graph database for queries, updates and analysis. Following these goals, 1 investigate (1) a formal causal property graph model and the extension of graph query languages to support causal graph query operators and (2) the integration, consistency and (3) versioning of causal property graphs to assist domain experts in the execution of advanced causal processes.