Development of Real-time Soccer Match Comment Sentiment Analysis and Emoji Conversion System

This study proposes an innovative system designed to automatically analyze emotions in real-time comments during live sports broadcasts, particularly soccer matches, and convert them into appropriate emojis. This approach aims to overcome the limitations of viewer interaction and enable seamless emotional communication across language barriers. The system utilizes the KoBERT model, optimized for Korean natural language processing, for accurate text-based sentiment analysis. The real-time emoji conversion and display functionality is implemented using the React framework, and web socket technology is employed to achieve low-latency data processing in real-time. The model was trained using a large Korean conversation dataset and achieved an emotion classification accuracy of 71.12%. In terms of performance, the system can process 308.71 comments per second with an average latency of 37.52 milliseconds, proving its effectiveness in a live sports broadcast environment. The proposed system enhances the viewing experience by allowing users to express emotions intuitively, thus breaking the limitations of text-based communication. This system introduces a new paradigm for audience interaction in live sports broadcasts, promoting a more inclusive and engaging experience. Using emojis, which transcend language barriers, viewers can share their emotions without text, fostering real-time emotional exchange among a global audience. The research highlights the practical application of real-time text and sentiment analysis technologies. It provides a foundation for future enhancements, such as support for multiple languages and more advanced real-time data processing capabilities.