Hop-Constrained s-t Simple Path Enumeration on Large Dynamic Graphs

Jiujing Zhang, Shiyu Yang, Dian Ouyang, Fan Zhang, Xuemin Lin, Long Yuan

Abstract: Hop-constrained s-t simple path (k-st path) enumeration is a fundamental problem in graph databases and plays an important role in many real-world applications. Given a dynamic graph G, a source-target pair s-t, and a hop constraint k, we aim to efficiently compute k-st paths: list all simple paths within length k from s to t, and then continuously maintain the results against edge updates. Although the k-st path enumeration has been well studied in static setting, the existing works on static graphs cannot be applied or adapted to handle dynamic graphs efficiently. To address the challenges on dynamic computation, we propose a partial path-based index structure and an efficient enumeration algorithm based on the index. We also propose several well-designed techniques to efficiently maintain the index and locate the affected results with graph updates. Comprehensive experiments verify that our proposed CPE <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">update</inf> algorithm outperforms the state-of-the-art methods by up to 4 orders of magnitude on dynamic graphs. The experiment results also show that the time cost of our initialization step CPE <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">startup</inf> (including index construction) is similar to the state-of-the-art static method.