| dc.description.abstract | In the current digital era, the role of influencers, particularly on social media platforms, has grown significantly. One common feature utilized by businesses is the "followers" feature. However, this feature only groups influencers based on accounts that follow business accounts, necessitating more sophisticated methods to effectively identify influencers. This study proposes a novel method for determining top influencers by integrating graph coloring algorithms and the Leiden algorithm, referred to as the Leiden Coloring algorithm. This method leverages network analysis to identify patterns and relationships within large-scale datasets. First, the Leiden Coloring algorithm is used to partition the network into various communities. Once these communities are formed, the principles of graph coloring are applied, where each community is assigned a unique color, with the total number of colors not exceeding the number of communities. This process is complemented by degree centrality, which identifies nodes with high connectivity, signifying influencer positions. The method was validated using crawled data from the Twitter (X) social media platform with the keyword "GarudaIndonesia." Through three testing scenarios, this method successfully identified the top 10 accounts as key influencer marketers. Some accounts consistently appearing in the analysis results include IndonesiaGaruda, GarudaCares, and wandiseptian11. The Leiden Coloring method was compared to the Leiden method and demonstrated improved performance. The modularity value of the Leiden Coloring algorithm increased by 0.000033, processing time was reduced by up to 10,12 seconds, and the number of communities generated decreased by 2. | en_US |
