In-Memory Processing for Nearest User-Specified Group Search

Abstract

This paper presents a nearest user-specified group (NUG) search which called a clustered NN problem. Given a set of data points P and a query point q, NUG search finds the nearest subset c ⊂ P (|c| ≥ k) from q (called user-specified group) that satisfies given conditions. Motivated by the brute-force approach for NUG search requires O(|P|²) computational cost, we propose a faster algorithm to handle NUG problem with in-memory processing. We first define clustered objects above k as a user-specified group and the NUG search problem. Moreover, the proposed solution converts a NUG search problem to a graph formulation problem, and reduces processing cost with geometric-based heuristics. Our experimental results show that the efficiency and effectiveness of our proposed approach outperforms the conventional one.