Dealing with repeated objects in SNNagg

Abstract

Due to the constant technological advances and massive use of electronic devices, the amount of data generated has increased at a very high rate, leading to the urgent need to process larger amounts of data in less time. In order to be able to handle these large amounts of data, several techniques and algorithms have been developed in the area of knowledge discovery in databases, which process consists of several stages, including data mining that analyze vast amounts of data, identifying patterns, models or trends. Among the several data mining techniques, this work is focused in clustering spatial data with a density-based approach that uses the Shared Nearest Neighbor algorithm (SNN). SNN has shown several advantages when analyzing this type of data, identifying clusters of different sizes, shapes, and densities, and also dealing with noise. This paper presents and evaluates a new extension of SNN that is able to deal with repeated objects, creating aggregates that reduce the processing time required to cluster a given dataset, as repeated objects are excluded from the most time demanding step, which is associated with the identification of the k-nearest neighbors of a point. The proposed approach, SNNagg, was evaluated and the obtained results show that the processing time is reduced without compromising the quality of the obtained clusters.