Application for Monitoring Sentiment Analysis and Geolocation in Shared Vehicle Services
摘要
The research was oriented toward creating a cutting-edge application that uses sentiment analysis and geolocation to improve safety in shared vehicle services. The objectives outlined ranged from designing the application to share emotional states and locations in real-time to evaluate the accuracy of the incident detection system, comparing its effectiveness, and conducting pilot tests in real-life situations. Implementing the Convolutional Neural Network (CNN) model emerged as a core component, resulting in 74% accuracy in sentiment classification. The findings highlighted the system’s effectiveness in achieving remarkable recall rates, especially in categories such as calm (78%) and noise (81%). However, limitations were identified in the accuracy of certain emotional states, such as sadness (40%). Despite these limitations, the conclusions emphasize the success in achieving the objectives, highlighting the application’s usefulness for improving citizen security. Continuous improvements in the model, institutional collaborations to strengthen its integration into security systems, and educational campaigns for users focusing on ethics and privacy are proposed. In summary, the research provides a robust basis for future developments, highlighting the optimization and adaptability of the application in diverse urban environments.