Using graph databases to manage linked data
James M. Farrow · Wiley series in probability and statistics · 2015
By storing record data and the relationships between the records explicitly in a graph database a technology specifically designed to store 'natural' graph structures and query such information many shortcomings of the relational approach can be avoided, and new approaches to managing and exploiting linked data are enabled. This chapter compares a relational and a graph-based approach, and provides an overview of how linked record data can be managed using a graph database based on the experience gleaned in building such a system at SANT DataLink in South Australia: the Next Generation Linkage Management System (NGLMS). Whereas algorithms based on relational structures and comparisons have a complexity greater than linear complexity, the graph partitioning algorithms have a complexity that is linear on the number of nodes involved. The graph-based approach most naturally represents and allows the exploitation of which it is most desirable to capture the relationship between data items.