Constructing Effective Caches of Shortest Path Queries on Road Networks

Abstract

How to effectively utilize caching technology to support high-performance shortest path queries on road networks has become an important research problem since the popularization of locationbased services. Most of the existing methods design a statistical model to construct the cache with the shortest paths, and mainly consider the related query frequency obtained from the historical query logs. As a result, the paths in the cache may be concentrated in a few hot-spot areas of the road network, such as the prosperous areas of a city. Additionally, queries from remote areas may never be hit in the cache, which results in the cache hit ratio is not fully optimized. To address this problem, we define an effective model to quantify the effect of cached shortest paths; the model not only considers the related query frequency in the log but also the coverage of the paths on the road network. Then, we propose an efficient cache construction method to select the cached shortest paths from the query log. Furthermore, we design a hybrid index for the shortest paths in the cache, so that the efficiency of shortest path querying is significantly improved. Finally, we conduct comprehensive experimental studies on different datasets to verify that the cache hit ratio is greatly improved by using the proposed cache construction approaches.