A Multilingual Approach to Aspect-Based Sentiment Analysis on Gobis Suroboyo Application Reviews using LDA and SVM

The GOBIS application, developed by the Surabaya City Transportation Department, is a digital service designed to provide public transportation information and reduce traffic congestion. Despite having exceeded 100,000 downloads, the application has received numerous complaints from users, as reflected in the multilingual reviews on its platform. To ensure analytical consistency, this research focuses solely on reviews in Indonesian and English. Using Aspect-Based Sentiment Analysis (ABSA), this study employs Latent Dirichlet Allocation (LDA) for aspect identification and Support Vector Machine (SVM) for sentiment classification. The aim of this research is to determine the dominant aspects in user feedback and evaluate the effectiveness of the Support Vector Machine (SVM) model in classifying multilingual reviews. The research results show six main aspects that frequently appear in reviews, namely Application Features and Development, User Suggestions and Service Innovation, Error and Location Accuracy, Delay and Application Usability, Comfort and Service Quality, as well as Route Tracking and Vehicle Information. The Support Vector Machine (SVM) model, tested with 10-fold cross-validation, demonstrates consistent performance, achieving balanced metrics accuracy (74.16%), precision (73.76%), recall (73.54%), and F1-score (73.63%). This highlights its capability in handling multilingual sentiment analysis for application improvement.

Abstract licence: CC BY

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.