Elfanny, Alifia and Dr. lr. Dian Eka Ratnawati, S.Si.. M.Kom. (2024) Analisis Sentimen Berbasis Aspek Aplikasi Threads Menggunakan Data Ulasan Pengguna Pada Google Play Store Dengan Metode Support Vector Machine. Sarjana thesis, Universitas Brawijaya.
Abstract
Aplikasi Threads, yang dikembangkan oleh Meta Platform, menciptakan terobosan baru di panggung media sosial yang secara khusus ditujukan untuk pengguna Instagram. Keistimewaan aplikasi ini terletak dalam perbandingannya dengan Twitter, diakui sebagai pesaing serius bahkan diberi predikat "pembunuh" Twitter. Walaupun mendapatkan popularitas, tidak terhindar dari kritik dan keluhan. Meski perusahaan mengumpulkan berbagai ulasan untuk meningkatkan layanan, nilai rating mungkin tidak sepenuhnya mencerminkan esensi ulasan karena perbedaan persepsi. Menghadapi kompleksitas ini, penelitian ini menerapkan analisis sentimen berbasis aspek untuk mengidentifikasi akar permasalahan dan memberikan rekomendasi. Data diambil melalui web scraper menggunakan Google Play Scraper, kemudian dikategorikan ke dalam dua aspek utama, yaitu satisfaction dan error, dengan sentimen positif atau negatif. Proses selanjutnya melibatkan enam tahap pre-processing, seperti pembersihan data, case folding, normalisasi slang, stemming, tokenizing, dan filtering. Metode klasifikasi ulasan menggunakan Support Vector Machine dan pembobotan kata dengan Term Frequency-Inverse Document Frequency. Hasil klasifikasi diimplementasikan dalam aplikasi web untuk mengklasifikasikan ulasan berdasarkan aspek dan sentimen. Analisis mendalam menggunakan Root Cause Analysis digunakan untuk menggambarkan akar permasalahan dari ulasan bersentimen negatif melalui visualisasi data berupa word cloud. Dari hasil pengujian, terlihat performa tinggi dengan tingkat akurasi sekitar 93,39% untuk aspek dan 93,41% untuk sentimen, dengan f1-score sekitar 93,38% dan 93,40%.
English Abstract
Threads application, developed by Meta Platform, stands out as a tailored social media platform designed for Instagram enthusiasts. Positioned as a competitor to Twitter, it has earned the moniker of Twitter's potential "killer" due to its rising popularity. However, amidst its acclaim, Threads has not been immune to criticism and user complaints. In a proactive move to enhance user experience, Meta collected diverse reviews, recognizing the challenge of perceptual differences in conventional ratings. To unravel the nuances of feedback, a comprehensive study employed aspect-based sentiment analysis, delving into two primary categories: satisfaction and errors, encompassing positive and negative sentiments. The study utilized a meticulous six-stage pre-processing approach, involving data cleansing, case folding, slang normalization, stemming, tokenizing, and filtering. Employing a Support Vector Machine coupled with Term Frequency-Inverse Document Frequency for word weighting, the classification method achieved remarkable accuracy rates of approximately 93.39% for aspects and 93.41% for sentiments. The results were integrated into a web application that categorized reviews based on aspects and sentiments. Furthermore, an in-depth analysis, utilizing Root Cause Analysis and presenting findings through a visually informative word cloud, revealed the core issues embedded in negative sentiment reviews. This analytical prowess showcased high-performance metrics, with an accuracy rate of around 93.39% for aspects and 93.41% for sentiments, complemented by f1-scores hovering at approximately 93.38% and 93.40%.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | 0524150014 |
| Uncontrolled Keywords: | Aplikasi Threads, Analisis Sentimen Berbasis Aspek, Support Vector Machine, Term Frequency-Inverse Document Frequency, Root Cause Analysis-Threads Application, Aspect-Based Sentiment Analysis, Support Vector Machine, Term Frequency-Inverse Document Frequency, Root Cause Analysis |
| Divisions: | Fakultas Ilmu Komputer > Sistem Informasi |
| Depositing User: | Sugeng Moelyono |
| Date Deposited: | 13 Feb 2024 07:47 |
| Last Modified: | 13 Feb 2024 07:47 |
| URI: | http://repository.ub.ac.id/id/eprint/214292 |
![[thumbnail of DALAM MASA EMBARGO]](http://repository.ub.ac.id/style/images/fileicons/text.png) |
Text (DALAM MASA EMBARGO)
Alifia Elfanny.pdf Restricted to Registered users only Download (9MB) |
Actions (login required)
 |
View Item |