A Shortest Path Query Method Based on Tree Decomposition and Label Coverage

The shortest path query is one of core contents in graph theory study, various problems in the real world can be transformed into it to solve. With the increase of network scale, classic shortest path query algorithms cannot meet the query demand on large-scale graphs by reason of query efficiency, storage costs, etc. In order to solve above problems, we lucubrate on previous works, and propose a novel method based on tree decomposition and label coverage (TDLC-SP) which consists of two phases: offline pretreatment phase and online query phase. In the pretreatment phase, we propose a novel acceleration index method TDLC, it maps the graph into a tree, allocates minimum label coverage for each vertex to reduce redundant data storage and vertices traversal range; In the query phase, utilizing the TDLC index, query is completed by traversing the tree structure only once, it further improves the query efficiency. Experimental results on several real-world networks and synthetic datasets demonstrate the efficiency and effectiveness of the proposed methods.