Comparative Analysis of Louvain, Leiden, and Walktrap Algorithms for Community Detection in the Network of Indonesian Laws

This study evaluates the effectiveness of community detection algorithms in uncovering the structural framework of Indonesian legislation enacted from 2019 to 2024. Using social network analysis principles, community detection reveals complex interrelationships among entities, particularly those formed by cross-referenced legal provisions. The Louvain, Leiden, and Walktrap algorithms were selected for their distinct advantages in analyzing legal networks. Louvain is efficient for large-scale networks, Leiden ensures stable and high-quality partitions, and Walktrap is effective for detecting dense communities. The evaluation employed metrics such as modularity, conductance, coverage, and inter-cluster density to assess clustering performance. Results show that the Louvain algorithm outperformed others, excelling in modularity, coverage, and conductance. The Leiden algorithm showed a slight edge in modularity, while Walktrap excelled only in inter-cluster density. These findings highlight Louvain’s suitability for legal network analysis, providing valuable insights for improving community detection methods. This study contributes to the effective grouping of laws based on legal subject similarity, enhancing the understanding of legislative structures.