Conversation Graphs in Online Social Media

In online social media platforms, users can express their ideas by posting original content or adding comments and responses to existing posts, thus generating virtual discussions conversations. Studying these conversations is essential for understanding the communication behavior of users. This study proposes a novel approach retrieve popular patterns on using network-based analysis. The analysis consists two main stages: intent network generation. Users’ intention detected keyword-based categorization posts comments, integrated with classification through Naive Bayes Support Vector Machine algorithms uncategorized comments. A continuous human-in-the-loop further improves classification. To build understand among users, we conversation graphs starting from hierarchical structure directed multigraph network. experiments categorize 90% 98% accuracy real dataset. model then identifies relevant in terms shape content; finally determines relevance frequency patterns. Results show that most discussion obtained resemble real-life interactions communication.