• clustering
• data gathering
• differential evolution
• fcm
• genetic algorithm
• stigmergic architecture
• twitter

Social networks are increasingly being used as an information source, for example in case of natural disasters, sport finals or election results. Event-triggered information spread consists of tweets, posts, memes and more, which propagate in reaction to a specific event occurring.
In the literature, various methods have been proposed to extract meaningful data and to try to identify both peaks of interest concerning a specific term and interactions between groups of related terms.
However, the approach proposed in this thesis doesn’t follow a group of terms, but a particular event with all its related terms, extracted from Twitter. The most frequent words are processed using a stigmergic architecture to evaluate the similarity in time and then the results obtained are clustered using Fuzzy C-Means to identify the main episodes occurred during the event.
Tweets gathered during the 2019 Formula 1 British Grand Prix were analyzed as a case study for this method. The system was able not only to produce coherent clusters w.r.t. the main phases of the race, but also to prove itself capable of being applied to different types of events.