Tham khảo tài liệu 'advances in database technology- p9', công nghệ thông tin, cơ sở dữ liệu phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | 382 J. Zhang et al. Fig. 21. Cost vs. S O 16 7l O number of closest pairs retrieved k a I O accesses Fig. 22. Cost vs. k S 7 O made in CMTV00 . On the other hand the density of S does not affect significantly the accesses to the obstacle R-tree because high density leads to closer distance between the Euclidean pairs. The CPU time of the algorithm shown in Fig. 21b grows fast with 5 O because the dominant factor is the computation required for obtaining the Euclidean closest pairs as opposed to obstructed distances . Fig. 22 shows the cost of the algorithm with 5 7j O for different values of k. The page accesses for the entity R-trees caused by the Euclidean CP algorithm remain almost constant since the major cost occurs before the first pair is output . the k closest pairs are likely to be in the heap after the first Euclidean NN is found and are returned without extra IOs . The accesses to the obstacle R-tree and the CPU time however increase with k because more obstacles must be taken into account during the construction of the visibility graphs. 8 Conclusion This paper tackles spatial query processing in the presence of obstacles. Given a set of entities P and a set of polygonal obstacles O our aim is to answer spatial queries with respect to the obstructed distance metric which corresponds to the length of the Please purchase PDF Split-Merge on to remove this watermark Spatial Queries in the Presence of Obstacles 383 shortest path that connects them without passing through obstacles. This problem has numerous important applications in real life and several main memory algorithms have been proposed in Computational Geometry. Surprisingly there is no previous work for disk-resident datasets in the area of Spatial Databases. Combining techniques and algorithms from both aforementioned fields we propose an integrated framework that efficiently answers most types of spatial queries . range search nearest neighbors e-distance joins
