Awantina, Rachma (2023) Analisis Sentimen Dan Pemodelan Topik Pada Ulasan Pengguna Aplikasi Wattpad Di Google Play Store Menggunakan Naïve Bayes Classifier Dan Latent Dirichlet Allocation. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 10611910000030-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
10611910000030-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2025. Download (1MB) | Request a copy |
Abstract
Kemajuan era digital membantu aktivitas manusia karena segala kebutuhan tersedia dalam satu genggaman, termasuk membaca ataupun menulis. Berkembangnya aplikasi novel digital memudahkan pembaca mengakses novel melalui gadget serta platform self-publishing bagi penulis. Wattpad sebagai aplikasi novel digital terpopuler di Google Play Store dengan total unduhan lebih dari seratus juta menawarkan beragam fitur menarik yang bisa diunduh secara gratis. Hal tersebut tidak terlepas dari ulasan pengguna aplikasi Wattpad yang perlu diteliti karena membantu pihak developer untuk memperbaiki masalah terkait performa aplikasi yang biasanya tidak sesuai harapan ketika dijalankan. Tujuan penelitian yakni mengetahui hasil analisis sentimen menggunakan Naïve Bayes Classifier (NBC) dan pemodelan topik menggunakan Latent Dirichlet Allocation (LDA) pada ulasan pengguna aplikasi Wattpad. Data diambil dari Google Play Store dengan teknik web scraping. Analisis sentimen terbagi menjadi dua kelas, yaitu positif dan negatif yang memudahkan masyarakat dalam menilai kinerja Wattpad. Pemodelan topik tersusun atas beberapa kata yang mewakili topik utama sehingga masyarakat dapat memahami informasi terkini tentang Wattpad. Hasil penelitian menunjukkan kondisi imbalanced data karena karakteristik ulasan pengguna aplikasi Wattpad memiliki sentimen positif yang lebih tinggi daripada sentimen negatif. Metode NBC menghasilkan ketepatan klasifikasi dengan G-Mean, AUC, maupun F1-Score tertinggi pada data training. Selain itu, metode LDA menghasilkan jumlah topik sebanyak lima berdasarkan coherence score tertinggi. Topik tersebut antara lain keluhan pengguna terkait masalah dalam aplikasi Wattpad, cerita-cerita yang terdapat di Wattpad berkualitas, munculnya iklan saat mengakses Wattpad, pengalaman yang dirasakan selama menjadi pengguna Wattpad, serta opini penulis dan pembaca mengenai cerita berbayar.
==========================================================================================================================
The progress of the digital era helps human activities because all needs are available in one hand, including reading or writing. The development of digital novel applications makes it easier for readers to access novels through gadgets and self-publishing platforms for writers. Wattpad as the most popular digital novel application on the Google Play Store with a total download of more than one hundred million offers a variety of interesting features that can be downloaded for free. This is inseparable from user reviews of the Wattpad application which need to be examined because it helps developers to fix problems related to application performance which are usually not as expected when run. The research objective is to find out the results of sentiment analysis using the Naïve Bayes Classifier (NBC) and topic modeling using Latent Dirichlet Allocation (LDA) in Wattpad application user reviews. Data is taken from the Google Play Store with web scraping techniques. Sentiment analysis is divided into two classes, namely positive and negative which makes it easier for the public to assess Wattpad's performance. Topic modeling is composed of several words that represent the main topics so that people can understand the latest information about Wattpad. The results show that the data is imbalanced because the characteristics of the Wattpad application user reviews have higher positive sentiments than negative sentiments. The NBC method produces classification accuracy with the highest G-Mean, AUC, and F1-Score in the training data. In addition, the LDA method produces a total of five topics based on the highest coherence score. These topics include user complaints regarding problems in the Wattpad application, quality stories on Wattpad, the appearance of advertisements when accessing Wattpad, the experience felt while being a Wattpad user, as well as the opinions of writers and readers regarding paid stories.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Analisis Sentimen, Latent Dirichlet Allocation, Naïve Bayes Classifier, Pemodelan Topik, Wattpad. ============================================================ Latent Dirichlet Allocation, Naïve Bayes Classifier, Sentiment Analysis, Topic Modeling, Wattpad. |
| Subjects: | Q Science > Q Science (General) > Q325.5 Machine learning. |
| Divisions: | Faculty of Vocational > 49501-Business Statistics |
| Depositing User: | Rachma Awantina |
| Date Deposited: | 15 Feb 2023 10:35 |
| Last Modified: | 15 Feb 2023 10:35 |
| URI: | http://repository.its.ac.id/id/eprint/97335 |
Actions (login required)
 |
View Item |