Particle Swarm Optimization – Support Vector Machine (PSO-SVM) Algorithm for Journal Rank Classification

Abstract

Support Vector Machine (SVM) is a method with basic classification principles for data that can be separated linearly. As it developed, SVM is designed to work on non-linear problems by incorporating kernel concepts in high-dimensional space. The SVM method implemented in this study for classifying international journals using the SCImago Journal Rank (SJR) dataset. To overcome the disadvantages of SVM performance, the researchers used Particle Swarm Optimization (PSO) to optimize its performance. The purpose of using PSO is to get a better classification performance based on the parameters and functions of the kernel used and to approach the SJR classification system. The process includes normalizing and processing the data on the PSO, followed by implementation using the SVM method. The accuracy results obtained from PSO-SVM are 63.12% using Linear kernels. Based on these results, it assumed that PSO-SVM is still unable to optimize the approach in the SJR classification system if the system is 100% accurate.