| dc.description.abstract | MyPertamina is a digital-based application which is a combination of loyalty and e-payment. Apart from that, MyPertamina also monitors the distribution of subsidies appropriately to eligible people. With the increasing number of users using MyPertamina, reviews have become uncategorized and unstructured. Companies also need to understand users' views of the application, both its advantages and disadvantages, to make the application even better in the future. Therefore, an aspect-based sentiment analysis approach is needed to be able to extract useful information from user sentiment on each aspect contained in the review, so that it can be used as evaluation material for the development of the MyPertamina application. This research uses 5185 data taken from the Google Play Store using scrapping techniques. The data that has been obtained will then be labeled positive, neutral, or negative. Data that has been labeled will be cleaned in the preprocessing technique, then processed using BERT Word embedding and Bidirectional Long-Short Term Memory (BiLSTM) with a training data and test data proportion of 80:20, so that it can be used to look for sentiments and aspects of reviews user. There are five aspects used, namely systems, services, transactions, subsidy registration, and usefulness. After evaluation using a confusion matrix, the BERT word embedding and BiLSTM methods were able to obtain an accuracy of 92.86%. | en_US |
