조건이 있는 k-최근접 객체 질의 처리 방법

Recent applications of geographical information system(GIS) need queries with attribute conditions, keyword conditions, and spatial conditions(simply, DB-spatial-keyword queries).
k-Nearest neighbor queries with attribute and keyword conditions are a kind of the DB-spatial-keyword queries whose spatial condition is a k-nearest neighbor condition. A k-nearest neighbor query finds top-k spatial objects nearest to a given query point that satisfy attribute and keyword conditions. Major commercial/open-source systems that support k-nearest neighbor queries with attribute and keyword conditions can not process those queries efficiently since they do not take advantage of spatial locality of the results. In this paper, we propose three methods for efficient processing of k-nearest neighbor queries with attribute and keyword conditions. We then compare the performance of the proposed methods by the extensive experiments and show the advantages and disadvantages of performance of the proposed methods while varying the parameters. The results are the basis of the optimization of k-nearest neighbor query processing.