APPLICATIONS OF GRAPH THEORY IN SOCIAL NETWORK ANALYSIS

Abstract

Abstract: This study aims to understand how graph theory in social networks utilizing the complex algorithms like Graph Convolutional Networks (GCNs), PageRank, Betweenness Centrality, and Modularity. Using the described algorithms on the dataset of social network interactions, it was revealed that the proposed approaches are able to uncover community structures and intricate relations between individuals or subjects. It was concluded that the GCNs provided as high as 92% accuracy on node classification tasks and this rose above others. When it comes to the centrality of the identified influencers, PageRank presented an average of 0. 035 as in the case of Betweenness Centrality in which some nodes were deemed influential in the network with up to 0. 45. By applying the modularity-based community detection, the achieved modularity is 0. Sixty-two, while showing that the proposition holds dense communities of practice within the NIH network. They also showed the strength and relevance of the graph-theoretic perspectives, which can facilitate the comprehension of social networks. When comparing the strengths and weaknesses of the three algorithms, it is also apparent that their use in combination offers a detailed understanding of the evolving networks necessary for the use of marketing, healthcare, and urban planning. Concerning contributions, this research enriches the knowledge in the field of SNA particularly in the aspects that relate to the performance of different algorithms as well as in the applications of the approach.