Aulia, Jasmine (2026) Pendekatan Multitopic Stacking Ensemble Untuk Klasifikasi Ulasan Emosi Pada Media Sosial Berbasis LSA (Studi Kasus : Pt. Jawapos Media Televisi). Undergraduate thesis, UPN Veteran Jawa Timur.
|
Text (cover)
22083010074.-cover.pdf Download (3MB) | Preview |
|
|
Text (bab1)
22083010074.-bab1.pdf Download (208kB) | Preview |
|
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (bab2)
22083010074.-bab2.pdf Restricted to Repository staff only until 7 July 2029. Download (741kB) |
|
|
Text (bab3)
22083010074.-bab3.pdf Restricted to Repository staff only until 7 July 2029. Download (5MB) |
|
|
Text (bab4)
22083010074.-bab4.pdf Restricted to Repository staff only until 7 July 2029. Download (1MB) |
|
|
Text (bab5)
22083010074.-bab5.pdf Download (152kB) | Preview |
|
|
Text (daftar pustaka)
22083010074.-daftar pustaka.pdf Download (165kB) | Preview |
|
|
Text (lampiran)
22083010074.-lampiran.pdf Restricted to Repository staff only until 7 July 2029. Download (231kB) |
Abstract
The development of social media has encouraged changes in audience interaction patterns toward television content, including at PT Jawapos Media Televisi (JTV), a local television station in East Java. Audience comments on JTV’s social media platforms can serve as an important source of information for understanding audience responses and perceptions. However, these data are unstructured, use informal language, and contain various emotional expressions, making them difficult to analyze manually. This study aims to classify the emotions of JTV social media audience comments using the Stacking Ensemble method and to analyze dominant topics using Latent Semantic Analysis (LSA). The comment data were consolidated into three emotion classes, namely joy, neutral, and negative. The research stages include text preprocessing, testing preprocessing and n-gram combinations, feature extraction using Term Frequency-Inverse Document Frequency (TF-IDF), applying SMOTE to the training data, emotion classification modeling, model evaluation, and topic analysis using LSA. The Stacking Ensemble model was built by combining Random Forest, Support Vector Machine (SVM), and XGBoost as base learners, with Logistic Regression as the meta learner. Based on the evaluation results, the best model achieved an accuracy of 73.95% and an F1-score of 73.41%. In addition, LSA was used to identify dominant topic patterns in audience comments. As an output, this study also developed a Streamlit-based application to display prediction results, visualizations, and comment analysis interactively. Thus, this study contributes to the application of Natural Language Processing (NLP) in helping to understand social media audience perceptions more efficiently.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | T Technology > T Technology (General) | ||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Data Science | ||||||||||||
| Depositing User: | Jasmine Aulia | ||||||||||||
| Date Deposited: | 07 Jul 2026 07:19 | ||||||||||||
| Last Modified: | 07 Jul 2026 07:44 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/54252 |
Actions (login required)
 |
View Item |