Comparison between Louvain and Leiden Algorithm for Network Structure: A Review
Siti Haryanti Hairol Anuar, Zuraida Abal Abas, Norhazwani Mohd Yunos, Nurul Hafizah Mohd Zaki, Nurul Akmal Hashim, Mohd Fariddudin Mokhtar, Siti Azirah Asmai, Zaheera Zainal Abidin, Ahmad Fadzli Nizam
Abstract
Abstract In the real network, there must be a large and complex network. The solution to understand that kind of network structure is using the community detection algorithms. There are a lot of other algorithms out there to perform community detection. Each of the algorithms has its own advantages and disadvantages with different types and scale of complex network. The Louvain has been experimented that shows bad connected in community and disconnected when running the algorithm iteratively. In this paper, two algorithm based on agglomerative method (Louvain and Leiden) are introduced and reviewed. The concept and benefit are summarized in detail by comparison. Finally, the Leiden algorithm’s property is considered the latest and fastest algorithm than the Louvain algorithm. For the future, the comparison can help in choosing the best community detection algorithms even though these algorithms have different definitions of community.