Spatio-Textual Similarity Joins Using Variable Prefix Filtering and MBR Filtering

Abstract

Given a set of objects that carries textual and spatial information, a spatio-textual similarity join computes the pair of objects which are textually similar and spatially near. Huge amount of work has been done in considering spatial dimension but less work has been done in spatio-textual joins.
Now a days, due to the availability of GPS enabled devices, a huge amount of spatio-textual information is being generated which needs new methods to perform operation on this type of data. Here we study join operations for spatio-textual data and uses various optimization techniques such as efficient grid partitioning for spatial information, MBR-prefix technique for R-tree data structure, use of variable prefix-length of textual information and ordering in textual vectors on their TF-IDF value. We show the improvement of these optimizations in terms of running time as well as pruning of non-candidates.