Graph-Based Conversation Analysis in Social Media

Social media platforms offer their audience the possibility to reply posts through comments and reactions. This allows social users express ideas opinions on shared content, thus opening virtual discussions. Most studies networks have focused only user relationships or while ignoring valuable information hidden in digital conversations, terms of structure discussion relation between contents, which is essential for understanding online communication behavior. work proposes a graph-based framework assess shape conversations. The analysis was composed two main stages: intent network generation. Users’ intention detected using keyword-based classification, followed by implementation machine learning-based classification algorithms uncategorized comments. Afterwards, human-in-the-loop involved improving classification. To extract patterns among users, we built conversation graphs directed multigraph show our model at real-life experiments. first experiment used data from real challenge it able categorize 90% with 98% accuracy. second COVID vaccine-related discussions forums investigated stance sentiment understand how are affected parent discussion. Finally, most popular were mined interpreted. We see that dynamics obtained similar traditional activities.