Enhancing Community Detection with Weighted Louvain

Community detection in complex networks plays a crucial role in analyzing data structures. The Louvain algorithm is a widely used method for community detection; however, it can be improved by incorporating edge weight information. In this study, we introduce a weighted version of the Louvain algorithm and compare it with the standard Louvain, Leiden, and Girvan-Newman methods. Experiments were conducted on six networks, including three small graphs (fewer than 500 nodes) and three large graphs (approximately 2500 nodes). The results show that the weighted Louvain algorithm not only achieves higher modularity than the original Louvain and other methods, but also reduces the execution time compared to the original Louvain. This improvement is attributed to leveraging edge weight information, which leads to faster convergence and more precise network partitioning.