Fast Nearest Neighbor Search with Keywords Using IR2-Tree

Abstract

Conventional abstraction queries, like vary search and nearest neighbor retrieval, involve alone conditions on objects geometric properties. Today, many trendy applications concern novel varieties of queries that aim to go looking out objects satisfying every a abstraction predicate, and a predicate on their associated texts. As associate example, instead of considering all the restaurants, a nearest neighbor question would instead provoke the eating place that is the nearest among those whose menus contain asteak, ˆ spaghetti, brandyaˆ all at identical time. Currently, the best answer to such queries depends on the IR2-tree, which, as shown throughout this paper, contains many deficiencies that seriously impact its efficiency. motivated by this, It tend to develop a latest access methodology called the abstraction inverted index that extends the traditional inverted index to subsume f-dimensional info, and comes with algorithms that will answer nearest neighbor queries with keywords in real time. As verified by experiments, the projected techniques trounce the IR2-tree in question latent amount considerably, generally by a part of orders of magnitude.