Top-k shortest paths in directed labeled multigraphs

Abstract

A top-k shortest path algorithm finds the k shortest paths of a given graph ordered by length. Interpreting graphs as RDF may lead to additional constraints, such as special loop restrictions or path patterns. Thus, traditional algorithms such as the ones by Dijkstra, Yen or Eppstein cannot be applied without further ado. We therefore implemented a solution method based on Eppstein’s algorithm which is thoroughly discussed in this paper. Using this method we were able to solve all tasks of the ESWC 2016 Top-k Shortest Path Challenge while achieving only moderate overhead compared to the original version. However, we also identified some potential for improvements. Additionally, a concept for embedding our algorithm into a SPARQL endpoint is provided.