Anatomy of Graph Matching based on an XQuery and RDF Implementation

Abstract

Graphs are becoming one of the most popular data modeling paradigms since they are able to model complex relationships that cannot be easily captured using traditional data models. One of the major tasks of graph management is graph matching, which aims to find all of the subgraphs in a data graph that match a query graph. In the literature, proposals in this context are classified into two different categories: graph-at-a-time, which process the whole query graph at the same time, and vertex-at-a-time, which process a single vertex of the query graph at the same time. In this paper, we propose a new vertex-at-a-time proposal that is based on graphlets, each of which comprises a vertex of a graph, all of the immediate neighbors of that vertex, and all of the edges that relate those neighbors. Furthermore, we also use the concept of minimum hub covers, each of which comprises a subset of vertices in the query graph that account for all of the edges in that graph. We present the algorithms of our proposal and describe an implementation based on XQuery and RDF. Our evaluation results show that our proposal is appealing to perform graph matching.