Akademik Veri Yönetim Sistemi
3rd Serbian International Conference on Applied Artificial Intelligence, SICAAI 2024, Kragujevac, Sırbistan, 23 - 24 Mayıs 2024, cilt.1446 LNNS, ss.1-8, (Tam Metin Bildiri)
In recent years, significant research has been carried out to discover current events and the number of events from X’s data stream and to reveal the contextual dimension of events. One of the major challenges in this context is that most traditional methods have to estimate the number of events in order to contextually analyze events. Another problem is that some methods tend to detect events that often trigger a significant volume of communication. In this study, we propose a community-based event discovery system that reveals events by applying community detection on the graph representing the co-occurrence relations of n-grams that are frequently observed in X’s posts. We evaluate the effectiveness of the proposed system over a dataset used as a comparative dataset. Our system shows the success of discovering as many events as the number of existing events in an unsupervised manner. It also represents each event with words and multiple-word distributions, helping to analyze the subject of events contextually.