A hierarchical model for label constraint reachability computation

Abstract

Reachability query is a fundamental problem on networks currently emerging in various application domains. However, very little existing work has studied reachability with label constraints imposed on the edges/vertices of graphs. In this paper, we study the problem of answering Label-Constraint Reachability (LCR) in a labeled directed graph, i.e., whether a directed path confined to a given label set from a source query vertex to a target query vertex in the input graph exists. We propose a model called the Bipartite Hierarchical (BiHi) to speed up the LCR computation. The BiHi model adopts a representation organized as hierarchies of bipartitions, and a vertex-cover-based algorithm to label the to-most remaining subgraph. In addition, by extending the two-hop labeling, the BiHi model introduces a top-down hierarchical labeling strategy to compress the index. Extensive theoretical analyses and experimental investigations on both real-life and synthetic datasets demonstrate that our method can reduce the memory cost and construction time of index construction while maintaining high query processing efficiency.