Interval-Index: A Scalable and Fast Approach for Reachability Queries in Large Graphs

Abstract

Now more and more large graphs are available. One interesting problem is how to effectively find reachability between any vertex pairs in a very large graph. Multiple approaches have been proposed to answer reachability queries. However, most approaches only perform well on small graphs. Processing reachability queries on large graphs requires much storage and computation and still remains challenges. In this paper, we propose a scalable and fast indexing approach called Interval-Index, based on traversal tree-based partitioning and relabeling scheme. Our approach has several unique features: first, the traversal tree-based partitioning ensures access locality and parallelism in computation; second, continuous relabeling ensures fast querying and saves search space; third, we convert the entire graph database into a traversal tree graph on a smaller scale, to reach a compact storage structure. Finally, we run extensive experiments on synthetic graphs and real graphs with different sizes, and show that Interval-Index approach outperforms the state-of-the-art Feline in both storage size and the performance of query execution.