Article Per Year (5 Year)
p-Index From 2021 - 2026
0.23
P-Index
This Author published in this journals
All Journal Jurnal Inovasi Pendidikan dan Teknologi Informasi (JIPTI)
Buulolo, Ginna Anggriani
Unknown Affiliation
Computer Science & IT Education Other

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
ANALISIS SENTIMEN TERHADAP INFLUENCER MENGGUNAKAN LSTM DAN SVM (LINIER SVC) PADA KOMENTAR PLATFORM X Buulolo, Ginna Anggriani; Sanjaya, Ucta Pradema
Jurnal Inovasi Pendidikan dan Teknologi Informasi (JIPTI) Vol. 7 No. 1 (2026): Jurnal Inovasi Pendidikan dan Teknologi Informasi (JIPTI)
Publisher : Information Technology Education Department

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52060/da4wk642

Abstract

Social media platform X has become an important platform for influencers in shaping public opinion, but user comments are generally unstructured and use informal language, and have diverse sentiments that are difficult to analyze. This study aims to classify sentiment towards influencers on the X platform and compare the performance of the Deep Learning LSTM model and the Classic Machine Learning SVM model. Using 8,252 comments that have undergone text pre-processing. To overcome class imbalance, the BiLSTM model uses FastText embedding and Focal Loss, while the SVM model applies SMOTE oversampling. Performance is evaluated using accuracy and F1-macro, and the differences are tested using the McNemar test. The results show that BiLSTM achieves an accuracy of 0.68 and an F1-macro of 0.66, slightly higher than SVM (accuracy 0.67; F1-macro 0.65). However, the McNemar test indicated that the difference was not significant (p > 0.05). These findings imply that SVM remains a viable alternative for efficient short text classification on limited computational resources, while BiLSTM provides a slight improvement. The limitations of this study lie in the scope of data, which is restricted by keywords and the time range of collection, so generalization to other domains requires further study. The novelty of this study lies in the comprehensive comparison of the two approaches to informal influencer comments with different imbalance handling strategies. Keywords: FastText; Focal Loss; Influencer; LSTM; SVM.
Co-Authors Sanjaya, Ucta Pradema