Prediction algorithm based on web mining for multimedia objects in next-generation Digital Earth

New research questions related to big data have brought Digital Earth into a new data-intensive era. In this paper, we present a prediction algorithm based on data mining especially for multimedia objects in next-generation Digital Earth. We collect the useful data and preprocess them in client logs, mine the sequential pattern between the spatial objects and the multimedia objects, and obtain the association rules. The derived association rules can be used for prefetching some candidates in advance. The diverse experiment results show that our prefetching strategy based on mining web logs achieves higher efficiency than the other general or no prefetching ones.