Top-N Query Processing on Spatial Databases Considering Bi-chromatic Reverse k-Nearest Neighbors

• Main Authors: Li, Cha-Lun, 李嘉倫
• Other Authors: Chen, Arbee L.P.
• Format: Others
• Language: en_US
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/95465927913595844988

碩士 === 國立清華大學 === 資訊工程學系 === 98 === A reverse k-nearest neighbor (RkNN) query retrieves the data points regarding the query point as one of their corresponding k nearest neighbors. A bi-chromatic reverse k-nearest neighbor (BRkNN) query is a variant of the RkNN query, considering two types of data. Given two types of data G and C, a BRkNN query regarding a data point q in G retrieves the data points from C that regard q as one of their corresponding k-nearest neighbors on G. Many existing approaches answer either the RkNN query or the BRkNN query. However, different from these approaches, we make the first attempt to propose a novel top-n query based on the concept of BRkNN queries in this thesis, which ranks the data points in G and retrieves the top-n ones according to the cardinalities of the corresponding BRkNN answer sets. For efficiently answering this top-n query, we construct the Voronoi Diagram of G to index the data points in G and C. The information related to the Voronoi Diagram of G can help to quickly compute the upper bounds of the cardinalities, thus efficiently pruning some candidate results. Moreover, based on an existing approach to answering the RkNN query and the characteristics of the Voronoi Diagram of G, we propose another method to find the candidate region regarding a BRkNN query, which tightens the corresponding search space. Finally, based on the triangle inequality, we also propose an efficient refinement algorithm for finding the exact BRkNN answer sets from the candidate regions. To evaluate our whole approach to answering the novel top-n query, it is compared with a naïve approach which applies a state-of-the-art algorithm for answering the BRkNN query to each data point in G. The experiment results reveal that our approach outperforms the naïve approach. Moreover, our approach to answering a single BRkNN query also outperforms this existing algorithm.dI