Towards Efficient Data Access Through Multiple Relationship in Graph-Structured Digital Archives

The research field of digital libraries mainly deals with data with graph structure. Graph database management systems (GDBMSs) are suitable for managing data in the digital library because the data size is large and its structure is complex. However, when performing a non-simple search or analysis on a graph, GDBMSs cannot avoid reaching already-scanned nodes from different starting nodes by repeatedly traversing edges such as property paths pattern in SPARQL. Therefore, when a GDBMS reaches high degree nodes, the number of graph traversals increases in proportion to the number of its adjacent nodes. Consequently, the cost of traversing multiple paths extremely increases affected by nodes connected enormous the number of edges in conventional GDBMSs. In this paper, we propose a data access approach by repeatedly traversing edges belonging to a specific relationship or anything one while distinguishing between high degree nodes and low degree ones. Finally, a result of our experiment indicated our approach can increase the speed of repeat traversals by a factor of a maximum of ten.