چكيده به لاتين
This study analyzes textual data from social media platforms and news outlets to understand public reactions and media representations of transportation events, focusing on safety issues in various transportation modes, including rail, road, air, and sea. Using advanced text mining techniques and natural language processing (NLP), the study aims to uncover patterns, sentiments, and biases in discourse related to transportation safety. The analysis identifies key topics and sentiments expressed in user-generated content and news articles, and examines the temporal distribution of these topics to identify emerging trends and shifts in public opinion and media representation.
The extracted data from platforms such as X (formerly Twitter), Instagram, and news media cover five-year, three-month, and one-year periods, respectively. The data were processed using natural language processing techniques. ParsBERT was employed for sentiment analysis, and the data were preprocessed with the Hazm tool, which included tokenization, normalization, and the removal of stop words. Visualization techniques such as word clouds were used to display key patterns. In sentiment analysis, ParsBERT results showed that most posts were neutral. Specifically, on X, 88% of posts were neutral and 12% negative; on Instagram, 80.7% of posts were neutral and 19.3% negative; and in the media, 88.1% were neutral and 11.9% negative. In contrast, in simple sentiment analysis, the proportion of positive posts was significantly higher.
This study develops a framework for using social media analysis to enhance situational awareness and analyze user sentiments on transportation-related issues discussed on platforms such as X, Instagram, and news media. Techniques like Latent Dirichlet Allocation (LDA) were used to categorize topics related to transportation challenges. Through case studies, this research demonstrates the effectiveness of the proposed framework in providing timely and relevant information for identifying and responding to transportation challenges. The results of topic modeling indicate that each communication platform addresses transportation issues in a distinct way. On X, users react quickly to everyday issues such as accidents and air pollution, using the platform to express immediate concerns and analyze ongoing events. Instagram, with its focus on visual and emotional content, is more dedicated to sharing users' personal experiences with transportation problems, fostering an emotional connection between users and transportation services. In contrast, the media focuses on broader structural issues, such as infrastructure and air pollution, playing a key role in shaping public opinion and influencing transportation policy. These differences show that each platform can be a powerful tool for managing transportation challenges and providing public feedback. The topic modeling results clearly indicate that X offers much more real-time information about transportation-related challenges and issues compared to Instagram and news outlets.
The findings emphasize the importance of integrating real-time social and news data into traditional transportation monitoring systems. This integration offers a cost-effective approach to improving incident response, enhancing public safety, and reducing disruptions in transportation systems. By understanding the complex dynamics of public perception and media framing in transportation safety, this research provides valuable insights for transportation authorities and policymakers to improve situational awareness, communication strategies, and policy interventions.
