Spatio-textual similarity joins using variable prefix filtering
Vivek Gupta, Vikram Goyal · 2015
Spatio-textual similarity join retrieves a set of pairs of objects which are close spatially and have similar textual contents. Due to the high cost of matching complex objects, most of the algorithms proposed for join run in two phases. In the first step, a set of candidate pairs are selected which are then verified finally in the second step. There is always a trade-off between cost of final matching and cost of candidate pair generation. The goal of this paper is to prune out most of the non-candidates by using some low cost mechanism so that less cost is incurred while performing the final verification. In this paper, we study a couple of heuristics for their effectiveness for pruning out non-candidates in the context of spatial-textual similarity join. We experimented on two real life datasets, Flickr and Foursquare, and find that using heuristics in a state of the art algorithm improves the performance.