IMPLEMENTASI INDOBERT DALAM ANALISIS SENTIMEN DATA ULASAN APLIKASI PLN MOBILE
Total View This Week0
Total View This Week0
Institusion
Institut Teknologi Perusahaan Listrik Negara
Author
TUA, SAHAT GOHI
Asri, Yessy
kuswardani, Dwina
Asri, Yessy
kuswardani, Dwina
Subject
Teknik Informatika
Datestamp
2023-05-31 06:46:24
Abstract :
The PLN Mobile application is a digital application created by PT PLN (Persero) with the aim of providing electricity services through a mobile application. Reviews on Google Playstore have a rating of 1 to 5, but users often give a rating that does not match the review so that this does not adequately describe the quality of the application. The number of reviews or data reviews on the PLN Mobile application is so large that it takes time and time to read in its entirety. To determine public opinion, a classification system is applied. This study aims to analyze the review sentiment data of the PLN mobile application with the Indonesian language model, IndoBERT. The methods used are deep learning and sentiment analysis. The data used is 1000 data from a population of 67949 (January ? June 2022). Sentiment analysis uses the IndoBERT model and the evaluation used is the confusion matrix. In this study using data from the Google-playstore web scrapping, and preprocessing the data, and labeling the data using TextBlob. The results of data labeling using TextBlob, resulted in 82.6% of data labeled neutral, 3.7% labeled negative and 13.7% labeled positive. The results of evaluating the Indobert classification using a ratio of 80:20 and calculating the model using epoch 3, learning-rate 0.003 , and a batch size of 32 produce an accuracy of 90%.
The PLN Mobile application is a digital application created by PT PLN (Persero) with the aim of providing electricity services through a mobile application. Reviews on Google Playstore have a rating of 1 to 5, but users often give a rating that does not match the review so that this does not adequately describe the quality of the application. The number of reviews or data reviews on the PLN Mobile application is so large that it takes time and time to read in its entirety. To determine public opinion, a classification system is applied. This study aims to analyze the review sentiment data of the PLN mobile application with the Indonesian language model, IndoBERT. The methods used are deep learning and sentiment analysis. The data used is 1000 data from a population of 67949 (January ? June 2022). Sentiment analysis uses the IndoBERT model and the evaluation used is the confusion matrix. In this study using data from the Google-playstore web scrapping, and preprocessing the data, and labeling the data using TextBlob. The results of data labeling using TextBlob, resulted in 82.6% of data labeled neutral, 3.7% labeled negative and 13.7% labeled positive. The results of evaluating the Indobert classification using a ratio of 80:20 and calculating the model using epoch 3, learning-rate 0.003 , and a batch size of 32 produce an accuracy of 90%.
Download
book
BibTex
Latex, Jabref
cloud_download
Original Resource
url resource Institution
