Analisis Performansi InSet dan VADER Lexicon pada Pelabelan Sentimen Menggunakan Algoritma SVM dan Optimasi PSO

Mahendra, Rafi Aditya (2025) Analisis Performansi InSet dan VADER Lexicon pada Pelabelan Sentimen Menggunakan Algoritma SVM dan Optimasi PSO. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 20081010208.-cover.pdf Download (2MB)
	Text (Bab 1) 20081010208.-bab1.pdf Download (441kB)
	Text (Bab 2) 20081010208.-bab2.pdf Restricted to Repository staff only until 19 February 2027. Download (711kB)
	Text (Bab 3) 20081010208.-bab3.pdf Restricted to Repository staff only until 19 February 2027. Download (1MB)
	Text (Bab 4) 20081010208.-bab4.pdf Restricted to Repository staff only until 19 February 2027. Download (2MB)
	Text (Bab 5) 20081010208.-bab5.pdf Download (417kB)
	Text (Daftar Pustaka) 20081010208.-daftarpustaka.pdf Download (511kB)

Abstract

Sentiment analysis is a study that analyzes people's opinions, sentiments, judgments, attitudes, evaluations, and emotions towards entities such as problems and events. One of the commonly used approaches in sentiment analysis is the lexicon-based method, where a text is analyzed based on a sentiment dictionary that groups words into positive or negative sentiment categories. Sentiment dictionaries that can be used include InSet and VADER Lexicon. A comparison of the accuracy generated from data with InSet and VADER Lexicon labeling is carried out with the aim of knowing which lexicon is more effective to use in the data labeling process. The dataset used in testing in this research is tweets on application X (Twitter) regarding rohingya refugees in Indonesia. In addition, data classification will be carried out using the Support Vector Machine (SVM) algorithm and optimization using the Particle Swarm Optimization (PSO) method to find the best parameters that can increase the accuracy value resulting from the testing process. This research also compares the accuracy value generated from data classification using SVM without optimization and data classification using SVM and optimization using PSO (SVM+PSO). The results of the data labeling process using InSet Lexicon are 886 positive sentiments, 2906 negative sentiments, and 353 neutral sentiments. Meanwhile, data labeling using VADER Lexicon produces 1353 positive sentiments, 1942 negative sentiments, and 850 neutral sentiments. The highest accuracy results are obtained on data with InSet Lexicon labeling and data classification using SVM and optimization using PSO which produces an accuracy of 90.7%.

Item Type:

Thesis (Undergraduate)

Contributors: