Discussion of graph reachability query with keyword and distance constraint

Abstract

Focusing on the problem of graph reachability query with keyword and distance constraint, a method of graph reachability query based on reference node embedding was proposed. Firstly, a very small part of representative reference nodes were selected from all nodes, shortest path distance between all nodes and these reference nodes were previously calculated. Next distance range was obtained based on the triangle inequality relation. Lastly according to the distance constraint and keyword in query condition, a conclusion of reachability could be drawn quickly. Comparative tests were done on data of social network and road network. Compared with Dijkstra algorithm, the proportion of drawing a reachability conclusion directly of the proposed algorithm is 84.6 % in the New York Road network, while 66.6 % in the Digital Bibliography & Library Project (DBLP) network, moreover, the running time are both shortened greatly, which is reduced by 87.4 % and 77.3 % respectively. The testing experimental data demonstrates that the computational complexity of online queries is reduced by using a small index cost through the method proposed in this paper, which is a good solution to graph reachability query with keyword and distance constraint suitable for weighted graph as well as unweighted graph, with wider value in application.