Comparative Analysis of Public Transportation Through Sentiment Analysis and Topic Modeling

Aksan A., Camgöz Akdağ H.

International Symposium for Production Research, ISPR 2023, Antalya, Turkey, 5 - 07 October 2023, pp.3-15 identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1007/978-3-031-53991-6_1
  • City: Antalya
  • Country: Turkey
  • Page Numbers: pp.3-15
  • Keywords: latent dirichlet allocation, public transportation, roberta, sentiment analysis, topic modeling
  • Istanbul Technical University Affiliated: Yes


In the modern digital age, social media emerges as a powerful tool for extracting various public opinions on a wide range of subjects. Among various platforms, Twitter stands as a comprehensive source of real-time, global sentiment analysis. This research uses Twitter’s expansive reach to delve into public opinions on public transportation in both the United Kingdom and India. Employing the RoBERTa (Robustly Optimized BERT Pretraining Approach), this study categorizes tweets into positive, neutral, and negative sentiments, offering a structured analysis of public feelings toward transportation services in these nations. Subsequent application of Latent Dirichlet Allocation (LDA) reveals the causes within the identified sentiments on specific areas of satisfaction and concern in both countries’ public transportation systems. The crucial insights obtained from this study aim to guide informed and strategic enhancements in the public transportation sectors of both nations, pinpointing exact areas that demand attention and improvement.