Удосконалення алгоритму Лейдена для виявлення інфлюенсерів

An influencer is someone who has the ability to persuade a large number of people to take specific actions, regardless of space or time. The role of influencers, especially on social media platforms, has grown significantly. One common feature utilized by businesses today is follower grouping. However, this feature is limited to identifying influencers based solely on mutual followership, highlighting the need for a more advanced approach to influencer detection. This study proposes a new method that integrates the Leiden coloring algorithm with Degree centrality for influencer detection. This approach employs network analysis to identify patterns and relationships within large-scale datasets. First, the Leiden coloring algorithm partitions the network into various communities, which are considered potential influencer communities. Degree centrality then enhances this process by identifying highly connected nodes, which are indicative of influencers. The proposed method is validated using crawled data from Twitter (X) with the keyword “GarudaIndonesia”. The data collection process was carried out using Tweet Harvest, resulting in a dataset of 22,623 rows. The dataset was tested across three scenarios: the first with 1,000 rows, the second with 2,000 rows, and the third with 5,000 rows. The proposed method was compared with the Louvain coloring method, showing an increase in the modularity value of the Leiden coloring algorithm by 0.0240. This increase demonstrates the Leiden method's ability to achieve more optimal network partitioning. Additionally, the Leiden coloring algorithm reduced the processing time by 14.85 seconds compared to the Louvain method, highlighting its faster performance. This is particularly important for applications requiring quick results, especially in big data analysis. Lastly, the Leiden algorithm reduced the number of communities by 1,149, producing a simpler and more organized community structure, which facilitates easier and more efficient analysis