Detecting Real-Time Correlated Simultaneous Events in Microblogs: The Case of Men’s Olympic Football

Although many predictive models have been designed to detect realtime events, there is still little progress in characterizing simultaneous events. Simultaneous events found in the sport domain can be used to understand how several correlated incidents occur at the same time to describe a specific phenomenon. We proposed a novel mechanism that uses Twitter messages in order to predict emotions associated with the final football match between Brazil and Germany in Rio Olympics 2016. Users’ opinions and their sentiments were extracted from the obtained tweets using the K-means clustering algorithm and the SentiStrength technique. We also applied the “Multi-label” classification technique in conjunction with the “Binary Relevance” (BR) method. The results showed that NaiveBayes was able to predict the match outcomes and related emotions with an accuracy value of 81% and a hamming loss value of 16%. This study provides a robust approach to successfully detect real-time events using social media platforms. It also helps football clubs to characterize matches during the time span of the game. Finally, the proposed method contributes to the decision-making process in the sport domain