Analisis Sentimen Berbasis Aspek pada Ulasan Aplikasi MyPertamina Menggunakan Convolutional Neural Network

Lestari, Noor Indah (2024) Analisis Sentimen Berbasis Aspek pada Ulasan Aplikasi MyPertamina Menggunakan Convolutional Neural Network. Diploma thesis, Institut Teknologi Sepuluh Nopember.

Text 2043201048-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (7MB) | Request a copy

Abstract

Kebijakan penggunaan aplikasi untuk mendapatkan BBM subsidi menimbulkan keluhan dari banyak sisi, mulai dari keharusan memiliki smartphone, jaringan internet yang belum merata di banyak wilayah, dan kestabilan aplikasi My Pertamina itu sendiri sehingga menimbulkan durasi antrean yang terkadang tidak terkendali. Selain itu, aplikasi ini semakin tidak menjamin penyaluran BBM bersubsidi untuk tepat sasaran karena banyak masyarakat yang tidak mempunyai smartphone yang justru merupakan masyarakat yang membutuhkan nantinya akan kesulitan mendapatkan BBM bersubsidi. Oleh karena itu, perlu dilakukan analisis sentimen pada ulasan dari pengguna agar dapat dijadikan sebagai evaluasi untuk mempertimbangkan berbagai perbaikan dan peningkatan. Namun, analisis sentimen hanya mampu menilai suatu ulasan cenderung positif atau negatif tanpa menunjukkan aspek apa yang menyertai sentimen ulasan tersebut sehingga diperlukan analisis lebih lanjut untuk dapat mengekstraksi aspek beserta sentimen yang diberikan pada ulasan aplikasi MyPertamina menggunakan Aspect-Based Sentiment Analysis (ABSA). Metode yang digunakan adalah Convolutional Neural Network (CNN) dengan menggabungkan ekstraksi fitur Word2Vec. Sentimen yang digunakan pada penelitian ini hanya berupa sentimen positif dan negatif. Sedangkan aspek yang digunakan, yaitu bug, subsidi, dan pembayaran. Aspect Classification Modelling didapatkan nilai akurasi sebesar 71% sedangkan Sentiment Classification Modelling didapatkan nilai akurasi sebesar 82%.
=================================================================================================================================
The policy of using applications to get subsidized fuel has given rise to complaints from many sides, starting from the requirement to have a smartphone, internet networks that are not evenly distributed in many areas, and the stability of the My Pertamina application itself, resulting in sometimes uncontrollable queue times. Apart from that, this application increasingly does not guarantee that the distribution of subsidized fuel is on target because many people who do not have smartphones are actually people who need it and will have difficulty getting subsidized fuel. Therefore, it is necessary to carry out sentiment analysis on user reviews so that they can be used as an evaluation to consider various improvements and enhancements. However, sentiment analysis is only able to assess whether a review tends to be positive or negative without showing what aspects accompany the review's sentiment, so further analysis is needed to be able to extract the aspects and sentiments given in the MyPertamina application review using Aspect-Based Sentiment Analysis (ABSA). The method used is a Convolutional Neural Network (CNN) combining Word2Vec feature extraction. The sentiments used in this research are only positive and negative sentiments. Meanwhile, the aspects used are bugs, subsidies and payments. Aspect Classification Modeling obtained an accuracy value of 71%, while Sentiment Classification Modeling obtained an accuracy value of 82%.

Item Type:	Thesis (Diploma)
Uncontrolled Keywords:	Aspect-Based Sentiment Analysis, Convolutional Neural Network, MyPertamina, Word2Vec.
Subjects:	H Social Sciences > HA Statistics H Social Sciences > HD Industries. Land use. Labor > HD108 Classification (Theory. Method. Relation to other subjects ) Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Vocational > 49501-Business Statistics
Depositing User:	Noor Indah Lestari
Date Deposited:	26 Jun 2024 02:51
Last Modified:	26 Jun 2024 02:51
URI:	http://repository.its.ac.id/id/eprint/108070

Actions (login required)

View Item