Analisis Sentimen Tweet Tentang UU Cipta Kerja Menggunakan Algoritma SVM Berbasis PSO

Abstract

Support Vector Machine (SVM) is one of the most widely used classification algorithms for sentiment analysis and has been shown to provide satisfactory performance. However, despite its advantages, the SVM algorithm still has weaknesses in selecting the right SVM parameters to optimize the performance. In this study, sentiment analysis was done with the use of data called tweets about Undang-Undang Cipta Kerja which reap many pros and cons by the people in Indonesia, especially the laborers. The classification method used in this study is the Support Vector Machine algorithm which is optimized using the Particle Swarm Optimization method for the SVM parameters selection in the hope of optimizing the performance generated by the SVM algorithm in sentiment analysis. The results of the study using 10 k-fold cross-validations using the SVM algorithm resulted in an accuracy of 92,99%, a precision of 93,24%, and a recall of 93%. Meanwhile, the SVM and PSO algorithms produce an accuracy of 95%, precision of 95,08%, and recall of 94,97%. The results show that the Particle Swarm Optimization method can overcome the weaknesses of the Support Vector Machine algorithm in the problem of parameter selection and has succeeded in improving the resulting performance where the SVM-PSO is more superior to SVM without optimization in sentiment analysis.

Improving Aspect-Based Sentiment Analysis for Hotel Reviews with Latent Dirichlet Allocation and Machine Learning Algorithms

“…This algorithm combines distinct features in a way that minimizes losses, reducing the number of features while retaining the most informative ones. When it comes to using Indonesian text, SVM was compared to LR, and it produced superior results compared to LR [19]. When the two algorithms are compared, it is evident that SVM is deterministic and logistic regression is probabilistic.…”

“…The data points closest to the hyperplane are commonly referred to as support vectors. Previous research has indicated that when combined with Particle Swarm Optimization (PSO), SVM can achieve an accuracy rate of 95% in the sentiment analysis process [19].…”