Sequential and MapReduce-based Algorithms for Constructing an In-Place Multidimensional Quad-Tree Index for Answering Fixed-Radius Nearest Neighbor Queries
Résumé
Answering fixed-radius nearest neighbor queries constitutes an important problem in many areas, ranging from geographic systems to similarity searching in object databases (e.g. image and video databases). The usual approach in order to efficiently answer such queries is to construct an index. In this paper we present algorithms for constructing a multidimensional quad-tree index. We start with well-known sequential algorithms and then adapt them to the MapReduce computation model, in order to be able to handle large amounts of data. In all the algorithms the objects are indexed in association with quad-tree cells (or nodes) which they intersect (plus possibly a few other nearby cells). When processing a query, multiple quad-tree cells may be searched in order to find the answer.
Fichier principal
Andreica_Tapus-In-Place_Indexing.pdf (6.08 Mo)
Télécharger le fichier
related_materials.zip (24.66 Mo)
Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Format : Autre
Loading...