Towards a Complete Direct Mapping From Relational Databases To Property Graphs

TL;DR

This paper presents a comprehensive and practical method for converting relational databases into property graphs, ensuring information, semantics, and query preservation, facilitating analytical tasks on complex data.

Contribution

It introduces the first complete and practical mapping process from relational databases to property graphs, including query translation to Cypher.

Findings

Preserves information, semantics, and queries during mapping.

Provides a practical translation of SQL queries to Cypher.

First solution to fully map relational data to property graphs.

Abstract

It is increasingly common to find complex data represented through the graph model. Contrary to relational models, graphs offer a high capacity for executing analytical tasks on complex data. Since a huge amount of data is still presented in terms of relational tables, it is necessary to understand how to translate this data into graphs. This paper proposes a complete mapping process that allows transforming any relational database (schema and instance) into a property graph database (schema and instance). Contrary to existing mappings, our solution preserves the three fundamental mapping properties, namely: information preservation, semantic preservation and query preservation. Moreover, we study mapping any SQL query into an equivalent Cypher query, which makes our solution practical. Existing solutions are either incomplete or based on non-practical query language. Thus, this work is the first complete and practical solution for mapping relations to graphs.