Historical traversals in native graph databases

Abstract

Since most graph data, such as data from social, citation and computer networks evolve over time, it is useful to be able to query their history. In this paper, we focus on supporting traversals of such graphs using a native graph database. We assume that we are given the history of an evolving graph as a sequence of graph snapshots representing the state of the graph at different time instances. We introduce models for storing such snapshots in the graph database and we propose algorithms for supporting various types of historical reachability and shortest path queries. Finally, we experimentally evaluate and compare the various models and algorithms using both real and synthetic datasets.