Bounded Paths for LCR Queries in Labeled Weighted Directed Graphs

Abstract

Most of the data in the Big Data era is semi-structured or unstructured that can be modeled as graphs where nodes could be objects and edges represent the relations among the objects. Given a source and destination vertex along with label-constraint set, the Label-Constraint Reachability(LCR) query finds the existence of a path between the source vertex and destination vertex within the label-constraint. The objective of the paper is to find the label-constrained paths efficiently bounded by cost. We extend and propose landmark based path indexing to compute bounded paths for LCR queries in graphs. It involves choosing a subset of nodes as landmark nodes, constructing an index that constitutes their reachable nodes and corresponding path information. For each non-landmark node, an additional index is constructed that constitutes the reachability to landmark nodes and their corresponding path information. In query processing, these indices are used to check for the reachability and find the bounded paths efficiently. Experiments were conducted on real graphs and benchmark synthetic datasets.