On efficient mutual nearest neighbor query processing in spatial databases

This paper studies a new form of nearest neighbor queries in spatial databases, namely, mutual nearest neighbor (MNN) search. Given a set D of objects and a query object q, an MNN query returns from D, the set of objects that are among the k 1 (P1) nearest neighbors (NNs) of q; meanwhile, have q as one of their k 2 (P1) NNs. Although MNN queries are useful in many applications involving decision making, data mining, and pattern recognition, it cannot be efficiently handled by existing spatial query processing approaches. In this paper, we present the first piece of work for tackling MNN queries efficiently. Our methods utilize a conventional data-partitioning index (e.g., R-tree, etc.) on the dataset, employ the state-of-the-art database techniques including best-first based k nearest neighbor (kNN) retrieval and reverse kNN search with TPL pruning, and make use of the advantages of batch processing and reusing technique. An extensive empirical study, based on experiments performed using both real and synthetic datasets, has been conducted to demonstrate the efficiency and effectiveness of our proposed algorithms under various experimental settings. This paper studies a new form of nearest neighbor (NN) queries, namely mutual nearest neighbor (MNN) search. Given a dataset D, a query point q, and two parameters k 1 and k 2 , an MNN query retrieves those objects p 2 D such that p 2 NN k1 (q) 1 and q 2 NN k2 (p), i.e., it requires each answer object 2 to be one of the k 1 nearest neighbors (NNs) to q and meanwhile has q as one of its k 2 NNs. Consequently, it considers not only the spatial proximity of the answer objects to q, but also the spatial proximity of q to the answer objects. In other words, the conventional NN query is asymmetric, while MNN retrieval is symmetric. Although it is well known that asymmetric NN search fits the requirements of lots of applications, there are still many other practical applications that require symmetric NN queries. Some real-life applications are presented as follows. Resource allocation. Consider that a logistic company A has six branches (labeled as p 1 , p 2 , p 3 , p 4 , p 5 , p 6), as shown in Fig. 1a. In order to guarantee the quality of service, company A assigns each branch two nearby branches as backup to provide necessary supports in …