High performance data access, integration and portability of Graph Databases

Many of today’s big data problems can be modelled and analysed using Graph Databases. Indeed companies such as Google and Facebook use Graph Databases to solve their mission critical problems. Graph databases allow relationships between data entities to be modelled much more effectively and efficiently than relational databases or other NoSQL databases.

There are presently two main types of Graph Databases: the ones based on Resource Description Framework (RDF) (e.g., Virtuoso) and ones based on Property Graphs (e.g .,Neo4j). Presently porting graph data between these two types of graphs databases are not straightforward since each uses a different data modelling technique and storage format. In addition, querying and processing are done differently too. This means accessing and processing of graph data in these different database types and performing integration tasks, as well as porting data between them, are complex and slow.

This research project would investigate:

* Data modelling and representation techniques that make porting data between the two database types easier and faster.
* Achieving high performance queries over federated graph databases (ie data integration).
* Effective indexing techniques that work for both database types.
* Scalability of these databases in large high performance computing platforms.
* Use of multi-mode high performance databases for storing graph data (eg MongoDB with its graph database support).