The query performance for tracing tags depends upon the distribution of tag trajectories in the data space. We examine a more efficient representation of tag trajectories by means of ordering the set of domain values. Our analysis shows that the order of Location IDentifiers (LIDs) can give a great impact on the efficiency of query processing. To find out the optimal ordering of LIDs, we propose a new LID proximity function to rearrange an arbitrary order of LIDs. This function enables logically adjacent tag trajectories, which are frequently accessed together, to be stored in close proximity on the disk. To determine the optimal sequence of LIDs in the domain, we also propose a reordering scheme of LIDs. The experimental results show that the proposed reordering scheme achieves better query processing than the previous methods of assigning LIDs.
展开▼
