ดัชนีสำหรับการตอบการสอบถามแบบสูงสุด k อันดับในระบบการแนะนำเชิงตำแหน่ง

Abstract

This research is an improvement for the top-k query answering on the UCLAF (User-centered Collaborative Location and Activity Filtering) model, part of location-based recommendation systems, in order to increase the efficiency. This work proposes tree-based index structures to improve the efficiency of the top-k location-based recommendation queries. They are the variation of multidimensional index structures i.e. aR-tree, bR-tree, and cR-tree. For each of the index, a special component is added into all non-leaf nodes as additional information for the top-k query purpose. In the tree traversal processes, the nodes are selected by verifying the additional component according to pre-specified conditions. Eventually, the nodes that pass the conditions are accessed while the others are pruned. From the experiments, the proposed work is much more efficient for top-k query answering for the UCLAF model. The result shows that when the number of data in UCLAF model is increased, the proposed method can decrease the execution for determining top-k results compared to the other methods. These results can be obtained from accessing only potential nodes and pruning the rest.