INFERRING COMMUNITY STRUCTURE THROUGH MAXIMUM DEGREE-BASED RANDOM WALK WITH RESTART

Community structure, a critical topological property of complex networks, has recently received extensive and in-depth attention from researchers. Recognizing the non -uniform degree distribution of nodes within network subgraphs, this paper presents a novel algorithm called MD-RWR (Maximum Degree -based Random Walk with Restart) for community detection in complex networks. The proposed algorithm not only excels at identifying overlapping communities but also enhances the objectivity and accuracy of the results. To evaluate its performance, the algorithm is tested on five real -world networks. The experimental results demonstrate its effectiveness in detecting communities, particularly when dealing with overlapping ones. Furthermore, the algorithm surpasses Walktrap, Infomap, LPA, and LPA-S algorithms in terms of modularity and NMI scores, while exhibiting faster execution time compared to these algorithms.