Leadership-oriented community detection: enhancing accuracy in social network analysis

In today’s age of digital transformation and growing reliance on the internet and social networks, analyzing these networks has become increasingly important. Social media platforms have a significant influence on public opinion, and online communities play a key role in driving social and political movements. Traditional methods of identifying communities within social networks often focus solely on measures such as modularity, coverage, or similarity. It is essential to note that in social networks, opinions often form around leaders and are shared among individuals with similar interests, as well as in large groups. Current approaches do not consider these factors. To address this, a new method, BLOCD (Balanced Leadership-Oriented Community Detection), is proposed to consider the aforementioned factors. The proposed method was evaluated using networks with known community divisions, achieving better Normalized Mutual Information (NMI) scores than existing methods. This approach not only improves community detection accuracy but also has potential applications in other areas, such as identifying outbreaks of propagation, where rumors are propagated around leaders and in larger local communities. This study highlights the significance of considering opinion dynamics and community balance in social network analysis to achieve more accurate and meaningful results.