Graph databases – Time to change!

Tue Kofoed-Enevoldsen

The presentation concerns itself with the theoretical and practical challenge of representing time (i.e. change) and perform historical analysis using graph database technology. Graph database technology has risen to become one of NoSQL’s most prominent members, offering both intuitive data modelling concepts as well as advantageous query capabilities. However the issue of representing and querying dynamic rather than static graphs is still an underexplored topic with no established industry standards and only very few apparent best practices. From the perspective of data warehousing this is clearly something that must be resolved, sooner rather than later. Based on work experience from within the Danish energy industry and Denmark´s tax authorities, the presentation brings forth multiple different options for modelling historical data sets as graph objects. In addition hereto a method for logically categorizing a graph object´s temporal aspects is proposed.

Tue works as a data and information architect for the Danish tax authorities. He and his architect colleagues are responsible for the design and implementation of SKATs new data warehouse in which NoSQL technology will play an important role. As of now he is leading the department´s exploration into the fitness of graph database technology for becoming a permanent component of the DW data architecture. Tue has a BA in Philosophy (specializing in systems theory) and a MD in Software Development Technology (specializing in Database Technology and Data Modelling). He wrote his master thesis in collaboration with the Danish company DONG ENERGY (now Ørsted), investigating the applicability of graph databases (specifically Neo4j) to storing and analyzing historical data stemming from DONG´s electricity grid (i.e. data concerning grid topology, volts and currents etc.).


Comments are closed.