An Effective Clustering-Based Prefetching Scheme for Spatial Databases System

Because of the characteristics of web spatial databases, restrictions on the applications of web spatial databases focus on accessing speed. The main solution for improving accessing speed is caching and prefetching. The existing prefetching schemes prefetch future spatial objects based on the distances of spatial locations rather than accessing contents. For the improvement of the accuracy of prefetching, we propose a clustering algorithm-clust Voronoi, which based on the web user logs. And then, we apply clust Voronoi to the prefetching scheme, named clust Pref. This prefetching scheme can automatically determine the number of clusters and give the real case of spatial databases accessing. Through a simulation environment, using a real data set, we draw a conclusion that this prefetch scheme can effectively improve the spatial databases accessing performance.