eprintid: 9584
rev_number: 7
eprint_status: archive
userid: 46
dir: disk0/00/00/95/84
datestamp: 2025-05-19 03:04:33
lastmod: 2025-05-19 03:04:33
status_changed: 2025-05-19 03:04:33
type: thesis
metadata_visibility: show
creators_name: Solihin, Abdu Muhammad
creators_NPM: 065117127
contributors_type: http://www.loc.gov/loc.terms/relators/THS
contributors_type: http://www.loc.gov/loc.terms/relators/THS
contributors_name: Awaliyah Zuraiyah, Tjut
contributors_name: Suhartini, Dini
contributors_NIDN: 0409017301
corp_creators: Universitas Pakuan
corp_creators: Fakultas Matematika dan Ilmu Pnegetahuan Alam
corp_creators: Program Studi Ilmu Komputer
title: Comparison of Sentiment Analysis on Digital Platform
Application Reviews Using Naive Bayes
and Support Vector Machine
ispublished: pub
subjects: QK
divisions: sch_ecs
full_text_status: restricted
abstract: Comparison of Sentiment Analysis on Digital Platform
Application Reviews Using Naive Bayes
and Support Vector Machine
Abdu Muhammad Solihin1*
, Tjut Awaliyah Zuraiyah 2
, Dini Suhartini 3
1,2,3 Department of Computer Science, Faculty of Mathematics and Natural Sciences, Pakuan
University, Bogor, West Java, 16143, Indonesia
Abstract
The development of digital technology has changed how users interact with various services through
applications. User reviews on digital platform applications have become an important source of
information for understanding user sentiment and experience. This research analyzes user sentiment
towards digital platform applications and compares the performance of the Naive Bayes and Support
Vector Machine (SVM) algorithms. The data used is user reviews of the TipTip application from the
Google Play Store and App Store. The research methodology includes preprocessing, Term FrequencyInverse Document Frequency (TF-IDF) weighting, and classification using Naive Bayes and SVM with
Linear, RBF, Polynomial, and Sigmoid kernels. To address data imbalance, the Synthetic Minority
Oversampling Technique (SMOTE) is used. The results show that SVM with the RBF kernel achieves
the highest accuracy (98%), while Naive Bayes reaches 90%. The analysis also reveals a general trend of
positive user sentiment towards the application. This study demonstrates the effectiveness of SVM,
especially with the RBF kernel, in sentiment analysis of application reviews and highlights the impact of
SMOTE in handling data imbalance.
Keywords: Sentiment Analysis; Naive Bayes; Support Vector Machine; SMOTE; Application Reviews
date: 2024-08-08
date_type: published
pages: 20
institution: Universitas Pakuan
department: Fakultas Matematika dan Pengetahuan Alam
thesis_type: Skripsi
thesis_name: Sarjana
citation:   Solihin, Abdu Muhammad  (2024) Comparison of Sentiment Analysis on Digital Platform Application Reviews Using Naive Bayes and Support Vector Machine.  Skripsi thesis, Universitas Pakuan.   
document_url: http://eprints.unpak.ac.id/9584/1/Skripsi_Abdu%20Muhammad%20Solihin_065117127_Ilmu%20Komputer.pdf