Improving the Clustering Search heuristic: An application to cartographic labeling

The use of hybrid metaheuristics is a good approach to improve the quality and efficiency of metaheuristics. This paper presents a hybrid method based on Clustering Search (CS). CS seeks to combine metaheuristics and heuristics for local search, intensifying the search on regions of the search space which are considered promising. We propose a more efficient way to detect promising regions, based on the clustering techniques of Density-based spatial clustering of applications with noise (DBSCAN), Label-propagation (LP), and Natural Group Identification (NGI) algorithms. This proposal is called Density Clustering Search (DCS). To analyze this new approach, we propose to solve a combinatorial optimization problem with many practical applications, the Point Feature Cartographic Label Placement (PFCLP). The PFCLP attempts to locate identifiers (labels) of regions on a map without damaging legibility. The computational tests used instances taken from the literature. The results were satisfactory for clusters made with LP and NGI, presenting better results than the classic CS, which indicates these methods are a good alternative for the improvement of this method. (C) 2018 Published by Elsevier B.V.