Examinando por Autor "Perez-Pelo, Sergio"

Mostrando 1 - 3 de 3

A fast variable neighborhood search approach for multi-objective community detection
(Applied Soft Computing (Elsevier), 2021-11) Perez-Pelo, Sergio; Sanchez-Oro, Jesús; Gonzalez-Pardo, Antonio; Duarte, Abraham
Community detection in social networks is becoming one of the key tasks in social network analysis, since it helps analyzing groups of users with similar interests. This task is also useful in different areas, such as biology (interactions of genes and proteins), psychology (diagnostic criteria), or criminology (fraud detection). This paper presents a metaheuristic approach based on Variable Neighborhood Search (VNS) which leverages the combination of quality and diversity of a constructive procedure inspired in Greedy Randomized Adaptative Search Procedure (GRASP) for detecting communities in social networks. In this work, the community detection problem is modeled as a bi-objective optimization problem, where the two objective functions to be optimized are the Negative Ratio Association (NRA) and Ratio Cut (RC), two objectives that have already been proven to be in conflict. To evaluate the quality of the obtained solutions, we use the Normalized Mutual Information (NMI) metric for the instances under evaluation whose optimal solution is known, and modularity for those in which the optimal solution is unknown. Furthermore, we use metrics widely used in multiobjective optimization community to evaluate solutions, such as coverage, ϵ-indicator, hypervolume, and inverted generational distance. The obtained results outperform the state-of-the-art method for community detection over a set of real-life instances in both, quality and computing time.
On the analysis of the influence of the evaluation metric in community detection over social networks
(MDPI, 2019) Perez-Pelo, Sergio; Sanchez-Oro, Jesus; Martin-Santamaria, Raul; Duarte, Abraham
Community detection in social networks is becoming one of the key tasks in social network analysis, since it helps with analyzing groups of users with similar interests. As a consequence, it is possible to detect radicalism or even reduce the size of the data to be analyzed, among other applications. This paper presents a metaheuristic approach based on Greedy Randomized Adaptive Search Procedure (GRASP) methodology for detecting communities in social networks. The community detection problem is modeled as an optimization problem, where the objective function to be optimized is the modularity of the network, a well-known metric in this scientific field. The results obtained outperform classical methods of community detection over a set of real-life instances with respect to the quality of the communities detected.
Strategic oscillation for the balanced minimum sum-of-squares clustering problem
(Elsevier, 2021) Martin-Santamaria, Raul; Sanchez-Oro, Jesus; Perez-Pelo, Sergio; Duarte, Abraham
In the age of connectivity, every person is constantly producing large amounts of data every minute: social networks, information about trips, work connections, etc. These data will only become useful information if we are able to analyze and extract the most relevant features from it, which depends on the field of analysis. This task is usually performed by clustering data into similar groups with the aim of finding similarities and differences among them. However, the vast amount of data available makes traditional analysis obsolete for real-life datasets. This paper addresses the problem of dividing a set of elements into a predefined number of equally-sized clusters. In order to do so, we propose a Strategic Oscillation approach combined with a Greedy Randomized Adaptive Search Procedure. The computational experiments section firstly tunes the parameters of the algorithm and studies the influence of the proposed strategies. Then, the best variant is compared with the current state-of-the-art method over the same set of instances. The obtained results show the superiority of the proposal using two different clustering metrics: MSE (Mean Square Error) and Davies-Bouldin index.