ADOPTING ASSOCIATION RULE LEARNING FOR DETECTING AND RANKING INFLUENTIAL USERS ON HASHTAGS

Social Media platforms are networks of relationships in which users are considered as nodes in a graph and relationships between users are considered edges. These graphs symbolize online groups or communities. It has been a challenge for researchers to study the evolution of such communities due to their rapid and dynamic development. The contribution of our work is twofold: (i) studying the effectiveness of a set of Direct and Indirect Influence Measures that identify the factors that urges users to participate in certain communities; (ii) proposing an Association Rule Mining technique to identify influential users on a dataset of a number of trendy hashtags. As per the first contribution, the effectiveness of the above listed measures, is studied via different types of users’ participation whether being highly retweeted, active posting users, influencing their followers to post, or even socially connected among other users. For the latter, the identified influential users are then ranked based on the different Influence Measures to study the effect of these measures on studying the evolution of the hashtags. The results of this study indicate the effectiveness of identifying influential users through Association Rule Learning and identifying the most effective Influence Measures for different hashtags and different categories.