Claim Missing Document
Check
Articles

Found 1 Documents
Search
Journal : Building of Informatics, Technology and Science

Dissatisfaction of a Mobile-Based Application from Different Platforms Using Naïve Bayes for Sentiment Analysis and LDA for Topic Modelling Ohoilulin, Anastasya; Inan, Dedi I; Juita, Ratna; Sanglise, Marlinda
Building of Informatics, Technology and Science (BITS) Vol 6 No 2 (2024): September 2024
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v6i2.5729

Abstract

A mobile application that is built and runs on different platforms, such as iOS (Apple App Store) and Android (Google Play Store), may not necessarily have the same user satisfaction (dissatisfaction) reviews understood by both user segments. This is due to, for example, the differences in the technology used, which ultimately result in different user behaviors. This can be observed from the average ratings on each platform, even though it is the same application. Therefore, this research aims to provide a foundation for the assumptions made. The case study used is the Satu Sehat mobile application, a widely utilized health service application. Text mining methods: sentiment analysis using Naive Bayes and topic modeling using Latent Dirichlet Allocation (LDA) were chosen due to their relevance to the research objectives. A total of 21,750 reviews from the Google Play Store and 7,350 reviews from the Apple App Store were collected using scraping techniques. The results showed that sentiment analysis model on negative sentiment in the Apple App Store excelled with a precision of 93%, recall of 93%, and F1-score of 95%, while in the Google Play Store it had a precision of 82%, recall of 87%, and F1-score of 85%. However, the performance of the positive sentiment model in the Apple App Store was very low, with a precision of 63%, recall of 33%, and F1-score of 43%, compared to the Google Play Store which had a precision of 78%, recall of 71%, and F1-score of 74%. This indicates that a higher level of dissatisfaction is observed in the Apple App Store compared to Android. These results are consistent with the average ratings of the application on both platforms. Topic modeling results, which presented 15 topics from each platform, showed similar common issues such as login, OTP verification, and data input errors on both platforms. However, reviews of the Satu Sehat running on the Apple tend to be more negative compared to the one of Android. Therefore, improving the application quality of the Apple platform is more expected to meet user expectations and enhancing the overall rating as in the Andrond one.