PERANCANGAN ALAT BANTU EKSTRAKSI TOPIK DATA ULASAN APLIKASI PEDULILINDUNGI MENGGUNAKAN METODE TEXT MINING

• Main Author: Wibowo Ciptono, William
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/67569
• Institution: Institut Teknologi Bandung

In order to deal with the pandemic, the Indonesian government developed the PeduliLindungi apps to prevent the spread of the Covid-19 virus. Initially, people responded positively toward the apps. However, the implementation process was rather dissatisfying for many parties because of many issues founded on the PeduliLindungi apps. This dissatisfaction occurs due to the difference of expectation between the community and the app’s developer. Based on these conditions, the developer needs to better understand people’s expectations. One of the media that can be used to understand people's expectations is application review data, since it contains problems and ideas for adding features from the customers. Despite that, due to the large number of user reviews, it is rather difficult for the developer to understand the needs of each individual user. Based on these problems, a tool was made to extract the topics from the PeduliLindungi apps user reviews using text mining technique. The reviews of PeduliLindungi went through a preparatory process consisting of tokenization, stop words & irrelevant words removal, and stemming process, before going into the modelling process using the Latent Dirichlet Allocation (LDA) model. Topic modeling using LDA produces topic groups and their relative prevalence scores. By obtaining topic groups and their relative prevalence scores, it is hoped that the developers will be able to get the user needs and the result can also be considered within the process of prioritizing the development points that will be carried out.