Analisis Sentimen Hashtag “Percuma Lapor Polisi” Pada Twitter Menggunakan Metode Support Vector Machine

Tamia, Kristina Natalia (2022) Analisis Sentimen Hashtag “Percuma Lapor Polisi” Pada Twitter Menggunakan Metode Support Vector Machine. Undergraduate Thesis thesis, Institut Telkom Telkom Purwokerto.

	Text 1. Cover.pdf Download (527kB)
	Text 3. Abstract.pdf Download (140kB)
	Text 2. Abstrak.pdf Download (147kB)
	Text 4. BAB I.pdf Download (325kB)
	Text 5. BAB II.pdf Download (412kB)
	Text 6. BAB III.pdf Download (367kB)
	Text 7. BAB IV.pdf Restricted to Registered users only Download (332kB) | Request a copy
	Text 8. BAB V.pdf Download (55kB)
	Text 9. Daftar Pustaka.pdf Download (386kB)
	Text 10. Lampiran.pdf Restricted to Registered users only Download (276kB) | Request a copy

Abstract

Social media is very close to human life, especially social media Twitter as a platform that is often used as a media campaign against an event. One of the events that got crowded and became trending topics on Twitter was the hashtag “Percuma Lapor Polisi”. Sentiment analysis is needed with the aim of knowing the picture of people's perceptions of this event. The method used in the sentiment analysis of this research is the Support Vector Machine (SVM). The stages of this research in general are data collection, pre-processing, labeling, word weighting, data sharing, classification using SVM, and evaluation of sentiment results. The research was conducted by two scenarios, where using the normalization process in pre-processing and not using the normalization process. In this study, Scenario 1 use the normalization process and Scenario 2 does not use the normalization process. The result of this research is the percentage distribution of positive sentiment in Scenario 1 is 50.23% and negative sentiment is 49.77%. Scenario 2 has a positive sentiment distribution percentage of 50.15% and a negative sentiment of 49.85%. The best of the SVM method is on Scenario 1 which is used normalization process with RBF kernel, the accuracy value is 83.96%. This is because the results of Scenario 1’s Pre-processing are better with normalization. Normalization makes the dataset more structured and affects the accuracy of SVM results. Keywords: Accuracy, Hashtag, Sentiment Analysis, Support Vector Machine, Twitter

Actions (login required)

View Item