Property Oriented Relational-To-Graph Database Conversion

Analysis of data stored in a graph enables the discovery of certain information that could be hard to see if the data were stored using some other model (e.g. relational). However, the vast majority of data in information systems today is stored in relational databases, which dominate the data management field over the last decades. In spite of the rise of NoSQL technologies, the development of new information systems is still mostly based on relational databases. Given the increasing awareness about the benefits of data analysis as well as current research interest in graph mining techniques, we aim to enable the usage of those techniques on relational data. In that regard, we propose a universal relational-to-graph data conversion algorithm which can be used in preparation of data to perform a graph mining analysis. Our approach leverages the property graph model which is mainly used by the graph databases, while maintaining the level of relational data clarity.